Background

Hamstring injuries are common in several sports, with an overall incidence of 1.2–4 injuries per 1000 h of athlete exposure [1,2,3]. In athletics and Gaelic football, they account for 17–21% of total injuries [3, 4] and it is suggested that approximately 22% of all football players sustain a hamstring injury each season [1]. Hamstring injuries result in an average time loss of 24 days [5] and, result in high cost for professional athletes and teams [6]. Furthermore, dancers exhibit a high incidence of muscle injuries [7]. The relevance of hamstring injuries in sports is therefore paramount.

A growing body of research has focused on hamstring injuries, specifically to identify risk factors [8,9,10] and to develop prevention and rehabilitation programmes [11,12,13,14,15]. However, there is no consensus on hamstring injury mechanism. Askling et al. [16] proposed two scenarios in which a hamstring injury may occur; during either high-speed running, or stretching movements [16]. The high-speed running type of injury typically affects the long head of the biceps femoris (BFlh) and has a shorter recovery time than the stretching type of injury, which commonly affects the semimembranosus (SM) [17,18,19]. The running type of injury is the most frequent [20, 21] and, in Australian football, 81% of hamstring injuries occur during sprinting, while kicking (stretching type) accounts for 19% of injuries [2]. In the literature, there are two theories on the mechanism of hamstring injuries sustained during running. One is based on the findings of Garret and Lieber et al. [22, 23], who believed that the hamstring is most susceptible to injury during active lengthening, typically observed during the late swing phase of the running gait cycle (Fig. 1) [24]. As a result, preventive studies have focused on eccentric strengthening, with, for example, the Nordic hamstring exercise, which is associated with a significantly lower injury incidence [25,26,27]. Mann et al. [28], however, proposed that hamstring injury occurs during the initial stance phase because of the large forces in opposing directions as the body is propelled forward over the touchdown point (Fig. 1). By defining the mechanism of injury, new preventive strategies can hopefully be created to help reduce the number of hamstring injuries and re-injuries among athletes and patients. The aim of this study was to investigate the hamstring injury mechanism in a systematic review.

Fig. 1
figure 1

The running gait cycle

Full size image

Methods

The methodology of this study was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [29].

Eligibility criteria

All the original studies that investigated the mechanism of hamstring injury or the biomechanical properties of the hamstrings were evaluated for eligibility. Hamstring injury was defined as a strain injury to the hamstring muscle group. Therefore, hamstring injuries with avulsion fractures were not considered for this systematic review. Studies were included if 1) they were written in English and; 2) conclusions were extrapolated on the mechanisms of hamstring injury. Literature reviews, systematic reviews, meta-analyses, conference abstracts, chapters from text-books and editorials were excluded, as well as studies where the full text could not be obtained.

Information sources and search

Electronic search

A systematic electronic literature search was conducted on 21 February 2017 using the PubMed (first available date), EMBASE (starting in 1974) and the Cochrane Library (first available date) databases by an expert in electronic searching. An updated search was performed on 30 May 2018 for the PubMed and Cochrane, while an EMBASE search was updated on 7 June 2018. A third search was carried out on 10 July 2019. For all databases, a similar search strategy was used, where the only differences were due to database configuration. The search strategies used a combination of Medical Subject Heading (MeSH) terms and “title/abstract” search. The search strategy consisted of “hamstring AND injury NOT anterior cruciate ligament”, including synonyms (Tables 5, 6, 7 in Appendix).

Other search methods

The reference lists of all studies read in full text were screened for potential studies not previously identified.

Data collection and analysis

Study selection

All titles and abstracts were read and studies of potential interest were reviewed in full text independently by two authors (Author 1 and Author 2) to decide on inclusion or exclusion. Disagreements were resolved through discussion with senior authors (Author 7 and Author 8).

Data collection process

The data extraction process was performed in duplicate (Author 1 and Author 2) using a piloted form of a Microsoft Excel (Microsoft, USA) spreadsheet and the following parameters were retrieved; author, year of publication, title, journal, number of study subjects, information on study subjects (age, sex) purpose, a detailed description of the methods used to assess injury mechanism (including important details such as the use of a treadmill or track, surface or needle electrodes, sampling rate if performing a video analysis, the use of reflective markers and/or force plates to measure ground reaction force), a summary of the results and the authors’ conclusions.

Data synthesis

The data synthesis was performed with a qualitative approach by gathering the authors’ results and conclusions, thereby excluding studies in which the hypothesised, suggested hamstring injury mechanism was not presented. Groups were created during the review process based on the common study methods used and different injury mechanisms reported. These groups are presented as stretch-related injuries, kinematic analysis, electromyograph-based kinematic analysis and strength-related injuries respectively.

Quality appraisal of included studies

The included studies were evaluated for their reporting quality using the Downs and Black Checklist [30] comprising 27 items. Ten of the items refer to the reporting of study results, three items refer to external validity, 13 items to internal validity and one item to power calculation. Since none of the included studies was interventional and only one study had comparative groups, a total of 16 items were used, while 11 were excluded from the qualitative analysis (items 4–5, 8, 13–15, 19, 21–24). Of the 16 items used, seven examined the reporting of information, two examined external validity, six investigated internal validity and one item was related to power calculation. Each item can be answered yes (1 point), no (0 points) and unable to determine (0 points), except item 27, which may yield up to five points depending on the power calculation. The maximum score on the modified Down and Blacks Checklist is 20. However, not all of the 16 included items were applicable to each individual study, as study methodologies differed. Two authors (Author 1 and Author 2) independently performed the quality appraisal and differences were resolved with discussion (Table 8 in Appendix).

Results

Study selection

The database search identified 318 studies from the Cochrane Library, 2053 from EMBASE and 1893 from PubMed, giving a total of 4264 studies. After the removal of the 1423 duplicates, the remaining 2841 studies were screened by abstract and title. Eligible studies underwent full text assessment and 21 studies were included in the final systematic review. During the full text assessment, 52 previously unidentified studies were identified from the reference lists (Fig. 2), of which five studies were eligible for inclusion [19, 28, 31,32,33].

Fig. 2
figure 2

The inclusion and exclusion of studies

Full size image

Risk of bias assessment

The quality appraisal with a modified version of the Downs and Black Checklist [30] resulted in a median (range) score of 8 (7–14) points of 20 possible. See Table 1 for full results.

Table 1 Scoring from the modified Downs and Black Checklist assessing risk of bias. Certain items were not applicable to all studies
Full size table

Characteristics of included studies

Of the 26 studies included, three investigated stretch-type hamstring injuries [19, 31, 45], 10 performed a kinematic analysis [28, 32, 35, 37, 39, 46, 47, 51,52,53], 10 additional studies performed a kinematic analysis combined with an electromyographic (EMG) analysis [33, 34, 36, 38, 41,42,43,44, 48, 54] and three analysed muscle strength [40, 49, 50]. The number of participants in the included studies ranged from one to 54 (total of 444 participants; some individuals included in more than one study) with an age range of 16–53 years.

Six studies analysed actual hamstring injuries [19, 31, 37, 45,46,47], one study compared previously injured and uninjured individuals [49], while 19 studies performed the analyses on uninjured individuals and estimated the hamstring injury mechanism [28, 32,33,34,35,36, 38,39,40,41,42,43,44, 48, 50,51,52,53,54]. A summary of the suggested hamstring injury mechanisms is presented in Table 2 and a comprehensive summary of the included studies can be found in Table 9 in Appendix.

Table 2 Summary of the suggested hamstring injury mechanisms and most injury-prone phase stratified by results and method used to investigate injury mechanism
Full size table

Stretch-related hamstring injuries

Three studies investigated hamstring injuries in dancers and water skiers and scored a median (range) of 10 points (8–11) out of 20 possible on the modified Downs and Black Checklist. The study populations ranged from 12 to 30 subjects aged 16–53 years who participated in interviews and clinical and magnetic resonance imaging (MRI)) examinations to determine the hamstring injury mechanism. All three studies reported that hamstring injuries occurred due to extensive hip flexion with a hyperextended knee [19, 31, 45]. In one study of dancers, the quadratus femoris and adductor magnus were injured simultaneously with the hamstrings [19].

Hamstring injury mechanism from kinematic analysis

Ten studies investigated the hamstrings through a kinematic analysis of study subjects aged 16–31 years with a median (range) score of 8 (7–9) of 20 possible on the modified Downs and Black Checklist. Nine of these studies were conducted on runners [28, 32, 37, 39, 46, 47, 51,52,53] and one on race walkers [35]. with study populations ranging from one to 20 participants. High-speed cameras and skin-placed markers on anatomic landmarks were most commonly used to study the injuries while the subjects ran on a treadmill or track. In four studies, a force plate was added to obtain additional information [35, 46, 47, 51]. One study measured BFlh dimensions using MRI images which were subsequently used in a simulation of hamstring injury mechanics [32]. Three studies were able to record a hamstring injury in real time [37, 46, 47]. However, two of these studies based their conclusions on data from the same study subject [46, 47]. Seven studies made estimations of where the hamstrings were at highest risk of injury [28, 32, 35, 39, 51,52,53].

Two studies reported that hamstring injuries occur during the early stance phase [28, 39], while running with a forward trunk lean [39]. In contrast, seven studies concluded that hamstring injuries occur during the swing phase [32, 35, 37, 46, 47, 52, 53] and one study concluded that both phases exhibit a risk of injury [51]. It was proposed that the late or terminal swing phase placed the hamstring muscles at the highest risk of injury (Table 3).

Table 3 Methodological characteristics of the kinematic studies
Full size table

Hamstring injury mechanism from kinematic and electromyographic analysis

Ten studies performed EMG-based kinematic analysis [33, 34, 36, 38, 41,42,43,44, 48, 54] measured with either surface or needle electrodes [33] and, in some cases, with additional force plates [36, 41, 48]. The modified Downs and Black Checklist yielded a total median (range) score of 8 (8–14) of 20 possible for these studies. Seven studies analysed runners [33, 34, 38, 41, 44, 48, 54], one study used race walkers [36], one evaluated volleyball players performing different jumping tasks [42] and one study compared muscle activity while standing on one leg with different trunk and pelvic positions in healthy volunteers [43]. The studies included recreational and high-level athletes with an age range of 18–53 years and consisted of seven to 30 individuals.

One study concluded that the risk of hamstring injury is greatest during the early stance phase [41], while five studies reported that injury occurred during the swing phase [33, 34, 36, 38, 48]. One study suggested that hamstring injury may occur during either the early stance phase or late swing phase [44], while another study reported that injury could occur during both the late stance and late swing phase (Fig. 1) [54].

One study reported that anterior trunk sway and contralateral pelvic drop while standing on one leg increased the load on the hamstrings [43], while another study reported that the hamstrings are at risk of injury during concentric, braking movements [42]. All conclusions were based on estimations of when the highest risk of hamstring injury occurs, i.e. no study included an actual hamstring injury (Table 4).

Table 4 Methodological characteristics of the kinematic studies with concomitant electromyographic analyses
Full size table

Strength-related hamstring injuries

Three studies investigated hamstring strength in football players aged 18–35 years [40, 49, 50] and scored a median (range) value of 10 points (9–11) of 20 possible on the modified Downs and Black Checklist. One study measured seated isokinetic strength in 20 football players prior to, during and after an exercise protocol set to simulate the muscle fatigue induced by a football game [40]. It was reported that hamstring injury was caused by lower eccentric strength due to fatigue [40]. Two studies used muscle functional magnetic resonance imaging (mfMRI) to compare metabolic activity before and after an eccentric hamstring exercise in previously uninjured and injured football players [49, 50]. One study reported that previously injured athletes had lower eccentric endurance of the hamstrings compared with uninjured athletes. It was proposed that the inferior hamstring endurance was a result of less economic muscle activation which may constitute a risk for injury [49]. One study performed an MRI analysis before and after an eccentric hamstring exercise and registered hamstring injuries for the following 1.5 seasons [50]. The results indicated that a greater contribution from the biceps femoris compared with the semitendinosus (ST) during an eccentric hamstring exercise correlates with first-time hamstring injuries. Re-injuries were associated with lower eccentric hamstring endurance [50].

Discussion

Across studies that investigated runners, the most commonly suggested injury mechanism was eccentric strain during the late swing phase of the running gait cycle. In a sub-group of hamstring injuries, the reviewed studies reported that the mechanism of hamstring injuries includes a simultaneous hip flexion and knee extension.

Stretch-related hamstring injuries

All the studies [19, 31, 45] of stretch-type injuries concluded that injuries occur due to extensive hip flexion with simultaneous knee extension. The study methods were similar, with a qualitative interview on the injury situation as the main source of information. In Australian football, a total of 19% of hamstring injuries occur during kicking [2], which is a typical stretch-type hamstring injury, given that the end of a kick exhibits both a flexed hip and extended knee position. In addition, Worth [55] suggested that trying to pick up a ball from the ground while running at full speed is the most common hamstring injury situation in Australian football. Picking up something from the ground may exhibit the same traits as the stretch-type hamstring injuries, further supporting this theory [55]. Notably, these studies analysed patients who had sustained hamstring injuries. However, since none of the hamstring injuries was observed by the researchers, the injury situations were recalled by the patient, thereby entailing a risk of bias. The findings relating to stretch-type hamstring injury should therefore be interpreted with caution.

Hamstring injuries during running

The majority of studies of hamstring injuries during running reported that the hamstrings are most prone to injury during the late swing phase as a result of eccentric loading. However, some studies reported that the hamstrings are most likely to be injured during the stance phase. It is pivotal to acknowledge that, in cases in which an accidental hamstring injury was recorded in real time, the authors concluded that the injury occurred during the late swing phase [37, 46, 47]. This information was concluded through the earliest sign of injury including neuromuscular latencies [37, 47] as well as examining hamstring length, force, velocity and negative work [46]. This is in line with the findings of a recent literature review which suggests that hamstring injury during the late swing phase occurs due to high levels of muscle excitation and muscle strain [56]. Interestingly, Mendiguchia et al. [57] were able to record a hamstring injury and, while no injury mechanism was reported, the authors stated that the injury occurred when the subject ran with an “abnormal increase in power compared with velocity qualities” [57].

One study concluded that a hamstring injury is most likely to occur during the stance phase when comparing a normal running technique with a technique in which the subjects run with a forward trunk lean [39]. These results are in line with the findings of Prior et al. [43], who reported that an anterior trunk sway during single leg stance, similar to positions which occur in pivoting sports, increased hamstring strain [43]. However, strain on the hamstring muscles and injury conditions during running with a forward trunk lean may differ from a normal running technique as the forwards trunk lean elongates the hamstring muscle causing more strain. Interestingly, a forward trunk lean had the greatest impact during the stance phase with the knee fully extended, similar to the stretch-type injury mechanism. The forward trunk lean can be caused by poor activation and control of the muscles of the core and hip, thereby increasing the strain and injury risk of the hamstrings [58,59,60,61]. For this reason, an in-depth knowledge of this type of injury is imperative and could be implemented in hamstring injury prevention and rehabilitation programmes, focusing on hip and core strengthening exercises in addition to traditional hamstring exercises.

Furthermore, static stretching may reduce both the ground reaction forces observed in the early stance phase and the strain on the BFlh during the late swing phase [44]. This results in subsequent reduced peak values of joint torque at the hip and knee and increased force productions of the biceps femoris at longer muscle lengths, which demonstrates that stretching may reduce the risk of hamstring injuries [44, 56]. These findings are of particular interest as preventive studies on the Nordic hamstring exercise which focuses on eccentric training have shown to reduce the risk of hamstring injuries [25,26,27]. The preventive effect of the Nordic hamstring exercise may be attributed to its ability to increase muscle fascicle length [62] as short hamstring fascicles are associated with an increased risk of a hamstring injury [63].

The results of a study of muscle activity during running and preventive exercises for the hamstrings suggested that the highest activity of the hamstrings occurs during the late swing phase [64], potentially associated with an increased risk of injury. On the other hand, Ono et al. [41] reported that, during the swing phase, the tensile forces in the ST exceed the forces in the BFlh, while the BFlh during the stance phase demonstrates higher forces. Since it is more common to injure the BFlh while running compared with the ST, the authors suggested that hamstring injury probably occurs during the stance phase [18]. In addition, the medial hamstrings are primarily loaded during the swing phase, where the lateral hamstrings are active throughout the entire gait cycle [65], which may help to explain why the ST is less injured, despite the high force [41].

In the light of these findings, several limitations need to be mentioned. There were only three case reports that studied recordings of a real-time hamstring injury [37, 46, 47] and the same study subject was used in two of the case reports [46, 47]. Furthermore, contextual conditions varied between studies, where, in some studies, the running analyses were performed on a treadmill [34, 37, 52] and had subjects running at a slow pace, which may not reflect the mechanism of hamstring injury. Since hamstring injuries commonly affect athletes playing various sports on grass fields, there is a lack of studies examining the injury mechanism in those conditions. The results in current literature may therefore prove difficult to apply to hamstring injuries sustained on grass. In addition, some studies performed a kinematic analysis without the use of an EMG which, it can be argued, only investigates hamstring lengthening and not active lengthening, i.e. eccentric contraction, as muscle activity is not measured.

In conclusion, hamstring injuries sustained while running or sprinting are estimated to occur during the late swing phase as a consequence of increased strain on the hamstring muscles. However, further research is needed to confirm these findings.

Strength-related hamstring injuries

There are inconclusive results from retrospective studies of hamstring strength in relation to the mechanism of injury. Fatigue was reported to reduce eccentric hamstring strength, which was suggested to increase the risk of a hamstring injury [40], while lower hamstring strength endurance was associated with a hamstring re-injury [50]. One study compared muscle activity in athletes with previously injured and uninjured hamstrings and reported that the previously injured athletes had inferior hamstring activation, which contributes to lower hamstring strength [49]. These findings are most probably related to risk factors for suffering a subsequent injury, which may in turn help to improve rehabilitation, rather than being related to the mechanism of hamstring injury [1, 5, 66].

Limitations

Most importantly, the majority of studies based their conclusions on estimations of the hamstring injury mechanism. Furthermore, the number of publications relating to the hamstring injury mechanism is limited and different methods have been used to assess the mechanism of injury. As a result, the included studies were allocated to groups defined by the study method and mechanism of injury. Each group included a limited number of studies with different methodological limitations which resulted in uncertainty about the results in this systematic review. In addition, a number of biomechanical studies were excluded, as no conclusions were drawn with regard to the hamstring injury mechanism. The extensive manual search of the reference lists of included studies helped to identify additional literature on the hamstring injury mechanism. However, the inclusion criterion of “conclusions were extrapolated by the authors with regard to the mechanisms of hamstring injury” may have introduced bias, as studies either estimated the mechanism of injury or retrospectively reviewed hamstring injuries and not an actual injury per se. Also, only studies written in English were set to be included but throughout the process of manually searching reference lists no studies were excluded for this reason.

The Downs and Black Checklist was deemed the most correct to determine the reporting quality of included studies, although it was not completely suited to the study designs included. The overall interpretation of reporting quality was low, with a risk of bias related primarily to study size and design, although there are no cut-offs or standardised methods for interpreting the modified version of the Downs and Black Checklist.

Conclusion

A stretch-type injury to the hamstrings is caused by extensive hip flexion with an extended knee. Hamstring injuries during sprinting are most likely to occur due to excessive muscle strain caused by eccentric contraction during the late swing phase of the running gait cycle.