Purpose: This study compared regional lumbar (upper and lower), pelvis, trunk, and lower limb kinematics between elite male adolescent players with and without a history of low back pain (LBP) during the kick and flat serves as well as regional lumbar mobility and serving kinematics relative to the end of range.
Methods: Seven players with a history of LBP and confirmed L4/L5 injury and 13 controls matched for age, height, mass, and performance underwent a three-dimensional motion analysis during serving trials and lumbar mobility assessments. Regional lumbar, pelvis, trunk, and lower limb kinematics were compared between pain/no pain and kick/flat serves using a series of 2 × 2 mixed-model ANOVA, with independent samples t-tests used to compare regional lumbar mobility between pain/no pain.
Results: The pain group had significantly reduced lower lumbar mobility in every plane of motion than the no pain group. The pain group demonstrated less right lower lumbar and pelvis/shoulder rotation, greater right pelvic tilt, earlier peak right knee extension velocity during the drive phase of the tennis serves, and greater lower lumbar and pelvis left rotation, upper lumbar left lateral flexion, and anterior pelvis tilt during the forward-swing phase. All players approached their lumbar end of range during the serve.
Conclusions: The results of this investigation suggest that a multidimensional LBP management and prevention strategy is required, including the assessment of regional spinal mobility, the lower limb and upper limb and spinal kinematics, and the integrated work between clinicians and coaches to adapt adverse technique.
1School of Physiotherapy and Curtin Health Innovation Research, Curtin University, Perth, AUSTRALIA; 2School of Sport Science, Exercise and Health, The University of Western Australia, Crawley, AUSTRALIA; and 3Sport Science and Medicine Unit, Tennis Australia, Melbourne, AUSTRALIA
Address for correspondence: Amity Campbell, School of Physiotherapy and Curtin Health Innovation Research, Curtin University, GPO Box U1987, Perth, WA 6845, Australia; E-mail: A.Campbell@curtin.edu.au.
Submitted for publication April 2013.
Accepted for publication July 2013.
Injuries to the lower back account for the greatest loss of playing time in adolescent male tennis players (5). The lower lumbar vertebrae (fourth and fifth lumbar vertebrae) of adolescent male players appear susceptible to pain and injury, with unpublished data indicating the presence of L4 or L5 stress reactions among 37.5% of the current Tennis Australia National Squad adolescent male players, who missed training owing to low back pain (LBP) in 2012. These data support the research of Aylas et al. (1), who reported that L5 stress reactions and disc/facet degeneration at the L4/L5 junction were the most common radiological abnormality in the lumbar spine of adolescent tennis players.
Speculation that stroke mechanics were related to the high rate of lower lumbar spinal pathology in tennis players (1,11,14,15) has received support through recent work linking greater lateral lumbar forces during the drive phase of the serve, combined with large vertical forces and simultaneous lateral flexion/extension movements, to a demonstrable history of LBP–pathology in an elite adolescent cohort (5). These players also exhibited reduced lumbar rotation during the drive phase of the serve and more rotation during the forward-swing phase than the controls (5). However, this analysis only considered the kinematics of a single lumbar region (i.e., L1–L5). However, the lower lumbar region (L3–L5) is known to be at greatest risk of injury in this population (1) and has demonstrated functional independence to the upper lumbar region (L1–L3) during dynamic movements, suggesting that regional differences in lumbar kinematics may be important (19,29). Therefore, the examination of upper and lower lumbar region kinematics independently in dynamic sports skills, like the tennis serve, is required.
The tennis serve requires the coordinated rotation of joints throughout the body (9,13). Rapid extension of the lower limb joints is known to be critical to the development of serve speed (24), which helps to explain its prominent role in serve instruction. Indeed, the influence of the lower limbs in spinal movements also has clinical significance and is therefore encouraged in LBP assessments and management strategies (18). Given the important link between the lower limbs/pelvis and spine, a better understanding of the interrelationship between lower lumbar and pelvis kinematics is needed during the service action. Further, shoulder/pelvis angles (or “separation angles”) have been documented to predict and/or differentiate players with and without a history of LBP in other sporting populations, including cricket fast bowlers (10,23). Together, these insights highlight the need for an improved understanding of the serve associated kinematics and their potential link to LBP. Given that the flat serve (FS) and the kick serve (KS) are the most popular first and second serves in male tennis, it is intuitive for regional lumbar kinematics to be explored in these two serve types.
Both the hyper- and the hypomobility of the lumbar region have been linked with LBP (21,22), yet lumbar mobility has not been comprehensively explored in adolescent tennis players. In this context, the range of motion assessments would permit the quantification of lumbar region serving kinematics relative to the end of range. Loading the spine when at end range flexion/extension is considered to place the passive spinal structures (e.g., vertebrae, intervertebral discs, and ligamentous structures) at greater risk of spinal injury than that in more neutral postures (21,22). As the lumbar region undergoes simultaneous extension/rotation and lateral flexion movements during the serve (5) and rotation range of motion is known to be limited at end range extension (3), quantifying the peak regional lumbar angles relative to end-range would help enlighten potential strain/injury mechanisms.
This study therefore aimed to compare regional lumbar (upper and lower), trunk, pelvis, and lower limb angles between elite male adolescent players with and without a history of LBP during the KS and the FS. Regional lumbar mobility and serving kinematics relative to the end of range were also compared between groups. The kinematic data presented in this article complement the kinetic data previously reported on the same subjects (5).
Adolescent male players from the current members of Tennis Australia’s National Academy and Australian Institute of Sport tennis programs were approached for recruitment (5). Participant eligibility and pain/no pain grouping were confirmed using a survey before data collection, resulting in 21 participants. The LBP (pain) group was defined as players with a history of at least two episodes of disabling LBP in the 12–15 months before data collection, which was aggravated by serving, required treatment, resulted in missed training/competition and was associated with verified lower lumbar spine (L4–L5) radiological abnormalities (injuries included stress reactions [n = 6] and spondylolysis [n = 1]) (5). Participants were excluded if within 3 wk of testing they had musculoskeletal injury or illness preventing maximum skill execution, had been on a modified training load, or at the time of testing indicated LBP during the serve. The control group (no pain) had no history of any severe LBP incident and were of similar height (mean ± SD; no pain = 178.0 ± 4.5 cm, pain = 180.4 ± 6.5 cm, P value = 0.628), mass (no pain = 69.4 ± 7.9 kg, pain = 67.7 ± 8.8 kg, P value = 0.672), and age (no pain = 15.6 ± 1.2 yr, pain = 16.6 ± 1.4 yr, P value = 0.136) to the pain group. The pain group had greater years playing experience than the no pain group (no pain = 9.0 ± 1.8 yr, pain = 11.0 ± 1.4 yr, P value = 0.029) (5).
The data collection was conducted at the Australian Institute of Sports’ Canberra biomechanics laboratory. Before data collection, each participant/guardian provided informed consent/assent to participate in this study, and relevant institutional ethical approval was granted (HR 144/2009). The VICON motion analysis system (Oxford Metrics Inc, Oxford, UK), operated at 250 Hz, was used to collect marker trajectory data.
After arrival to the laboratory, participant’s height and mass were recorded. Retroreflective markers (15-mm diameter) were fixed to the skin surface overlying specific trunk and lower limb anatomical landmarks (2,29).
The participant then completed a self-directed warm-up, followed by a series of trunk range of motion assessment trials according to established methods (19). Participants were asked to cross their arms and place their hands on their shoulders, then bend as far forward as possible until they reached end of range. This position was held for 5 s, and the peak flexion achieved was used to represent active end of range flexion. This task was repeated in lumbar extension as well as left/right lateral flexion and rotation (19). The participant then performed a series of serving trials from the deuce side of a full size replica tennis court. Serving trials were performed until three FS and three KS where the ball landed within a preset target area (1-m2 area bordering the T of the deuce service box) and was within 5 km·h−1 of their prerecorded maximal serve velocity were collected, as recorded by a stalker radar gun (Applied Concepts, Inc.).
VICON Nexus Software (Oxford Metrics Inc.) was used to treat data using standard biomechanical procedures, as previously detailed (5). After a residual analysis, the trajectories were filtered with a Woltering filter using a mean square error of 3, with ball–racquet impact treated in accordance with previously outlined methods (26,30). A customized mathematical model was used to calculate regional lumbar kinematics, using a right hand assumption (Table 1) (31).
A custom Labview program (National Instruments, Inc.) used racquet tip and ball position data to identify four previously established temporal events (ball release, first racquet high position, first racquet low point, and ball–racquet impact) (5,25). Peak racquet velocity was calculated from the resultant velocity of a marker placed on the tip of the racquet for each trial. Absolute peak upper and lower lumbar, shoulder/pelvis separation, and pelvis angles in each plane of motion were output for the drive (from racquet high point to racquet low point) and forward-swing (from racquet low point to impact) phases of each serving trial. Further, the upper and lower lumbar angle achieved during the active end range assessments was subtracted from the absolute peak lower lumbar and upper lumbar during each phase of the FS and KS. Therefore, a result of zero represented no difference between end of range and serve position, whereas a negative value indicated that the player had moved beyond their “end of range” during the serve. Finally, variables that have been previously reported to characterize the lower limb drive phase, including peak and time (% drive phase) of knee and hip flexion angle and extension angular velocities, were output (24).
Statistical analyses were performed using the Statistical Package for the Social Sciences (version 19; SPSS Inc., Chicago, IL). The range of motion differences between pain and no pain were compared using independent sample t-tests, with a critical alpha of <0.05 used to balance type 1 and 2 error. The average of each variable was calculated for each participant’s three KS and FS and then analyzed using a series of 2 × 2 mixed-model ANOVA to compare between pain/no pain (between participants) and KS/FS (within participants). Playing experience was incorporated as a covariate in all analyses. The interaction between pain status and serve type was analyzed and found not to be significant for any angle, permitting grouped statistical results to be presented. Alpha was adjusted a priori to ≤0.01 to account for the larger number of comparisons and maintain balance between the risk for type 1 and 2 errors. The sample size of 7 pain participants and 13 control subjects allowed 0.82 power to detect a 5° difference between groups, using an alpha of 0.01.
A comparison of the end of range in upper and lower lumbar regions revealed that the pain group had significantly less available lower lumbar spine movement in all planes than the no pain group (Fig. 1B), but with no differences in the upper lumbar spine range between the two groups (Fig. 1A). Specifically, the pain group had an average of 9.5° less extension range (95% confidence interval [CI] = 0.1–18.8, P = 0.047), 2.8° less left lateral flexion (95% CI = 0–5.5, P = 0.049), and 2.7° less right rotation (95% CI = 0–5.4, P = 0.047) than the no pain group. There was also a nonsignificant trend for the pain group to have less right lateral flexion range (mean difference (MD) = 4.1°, 95% CI = 0–8.2, P = 0.050) than the no pain group.
There was no significant effect of pain for the resultant racquet velocity of the FS (mean ± SD; pain = 48 ± 4 ms−1 and no pain = 49 ± 4 ms−1) or the KS (pain = 46 ± 3 and no pain = 46 ± 4). Although there was a trend for all players to use greater racquet resultant velocity when hitting the FS (MD = 2.6 ms−1, 95% CI = 0.4–5.4, P = 0.048).
The comparison of absolute movement of the trunk and pelvis regions between the pain and the no pain groups during the drive phase of the serves revealed that the pain group was characterized by significantly greater right lateral pelvic tilt (i.e., the pelvis was tilted such that the right side was higher than the left side) (MD = 6.4°, 95% CI = 2.1–10.7, P = 0.005), significantly less lower lumbar region right rotation (MD = 1.9°, 95% CI = 0.2–3.6, P = 0.02) and significantly smaller pelvis/shoulder separation angle (MD = 14°, 95% CI = 6.3–21.8, P < 0.001) than the no pain group (Table 2; Fig. 2). In contrast, during the forward-swing phase, the pain group used greater lower lumbar (MD = 2.1°, 95% CI = 0.7–3.6, P = 0.001) and pelvic (MD = 5.9°, 95% CI = 1.9–9.9, P = 0.005) rotation than the no pain group. There was also significantly greater peak upper lumbar left lateral flexion (MD = 4.8°, 95% CI = 1.8–7.7, P = 0.003) and anterior pelvis tilt (MD = 10.2, 95% CI = 5.2–15.3, P < 0.001) in the pain group during the forward swing.
Lower limb kinematics.
No significant pain and serve effects were observed for peak knee and hip flexion angles and hip extension angular velocity. However, the pain group demonstrated a trend for greater left knee extension velocities (MD = 83.8°·s−1, 95% CI = 5.3–162.3, P = 0.037) than the no pain group (Table 3). Further analysis of the temporal characteristics of the left and right peak knee extension velocities revealed that the pain group’s peak right knee extension occurred significantly earlier in the drive phase than the no pain group (MD = 12.5%, 95% CI = 3.3–21.6, P = 0.009).
Serve kinematics relative to end-range
The comparisons of the lumbar angles normalized to players’ end of range revealed no significant differences between groups or serve types (Table 4). It was evident that all players approached the end of range of the upper lumbar right rotation, the lower lumbar right lateral flexion and rotation during the drive phase, and the lower lumbar left lateral flexion and rotation end of range during the forward-swing phase of both serves.
This study found differences in lower lumbar mobility and kinematics during serving between elite adolescent males with and without LBP. The upper lumbar region kinematics were similar between groups, and there were no significant trunk and lower limb kinematic differences between the KS and the FS. These findings support the importance of investigating regional differences in the lower back, consistent with previous findings (19,29).
The players with a history of recurrent LBP and evidence of bone pathology at L4–L5 (stress reactions/spondylolysis) demonstrated reduced lower lumbar mobility compared with the no pain group in all planes of motion during active range of motion assessments. As this study was cross-sectional, it is not known whether these findings represent a cause or an effect of LBP. However, it is worth noting that six of the seven players in the pain group also reported an episode of LBP in the 12 months after the data collection, supporting the recurrent nature of their pain, whereas all players within the no pain group remained pain free. Unfortunately, as no electromyography data were attained, it is unclear whether the findings reflect a loss of mobility of the passive elements (e.g., ligamentous and bony structures and intervertebral discs) or a neuromuscular response stiffening the lower lumbar spine. Previous research has demonstrated that people with persistent lower lumbar pain commonly present with restricted movement of the lower lumbar spine and associated increased activation of the lower lumbar musculature (7,17,22,27,28). There is debate as to whether these responses are protective (adaptive) or provocative (maladaptive) (8,20), although there is some suggestion that increased lumbar stiffness is associated with recurrent LBP (12,17,27,28). Although increased neuromuscular stiffness may aid in the protection of spinal structures in an acute phase of a pain disorder (adaptive response), increased stiffness may increase back loading because of decreased shock absorption capacity leading to strain and pain over the longer term (maladaptive response) (20). Given no players reported pain at the time of testing, it is unlikely these findings reflect a protective pain response during the serve, although it is possible that pain memory mechanisms could play a role in augmenting increased neuromuscular tension (32).
The pain group players also demonstrated differences in the measured kinematics during the tennis serve, in comparison with the no pain group, although both groups achieved the same resultant racquet velocity. More specifically, the pain group demonstrated reduced lower lumbar, pelvis, and pelvis/shoulder right rotation, with their pelvis tilted laterally more to the right (or toward their racquet arm) and an earlier right knee extension velocity during the drive phase of the serves. Further to this, greater lower lumbar and pelvis left rotation as well as upper lumbar left lateral flexion characterized the forward swing of the pain group’s serves. Finally, although there were no differences between groups when the lower and upper lumbar region kinematics were normalized to the “end of range,” the drive phase in both serves was shown to require near end range lower lumbar extension, right lateral flexion and rotation movements, whereas lower lumbar left rotation and lateral flexion approached the end of range in the forward-swing phase.
Although we acknowledge the limitations of between group comparisons, the outlined results appear to highlight several potential factors related to the high rate of lower lumbar region injuries reported in the current population (1). First, the passive spinal structures, responsible for limiting movement when the lumbar vertebrae were positioned near end of range rotation (21,22), are likely to be placed under significant stress during serve (16). A loss of rotation mobility may result in earlier strain patterns occurring during serve in the pain group. Further, the unilateral pars interarticularis, a common location of pathology in this population (1), is known to be at greatest stress when under combinations of compression forces with lumbar extension, lumbar side flexion to the same side, and/or lumbar rotation to the opposite side. Therefore, the limitations reported in the pain group (particularly reduced rotation when at the end of range), coupled with the previously reported significantly higher compression and lateral flexion forces identified in players with LBP (5), are likely to be related to the high rate of lower lumbar spinal bone stress and pathology reported in this population. Lastly, the similarities evident between serve types highlight the potential for repeated strain, which serving places on the lower lumbar vertebrae.
Conclusion and implications
The results relating to both the serve kinematics and lower lumbar mobility suggest that a multilevel LBP management and prevention approach is required (20). This should include the assessment of regional spinal mobility during both planar and combined movements, lower limb and upper limb spinal kinematics. During this process, careful clinical examination is required to determine whether the reduced lower lumbar mobility of the pain subjects is a function of connective tissue structures or neuromuscular mechanisms (20). This provides a basis for integrated work between clinicians and coaches to enhance movement control and mobility and to adapt a technique to minimize stress on the lower lumbar spine. More specifically, the large differences in shoulder/pelvis rotation during the drive phase (14°) provide a target for preventative measures to reduce lumbar spine strain, as documented in other at risk populations (6,10,23). Further, the results support the hypothesis that the lower lumbar spine moves in synchrony with the pelvis (19), whereas the earlier right lower limb drive likely contributes to the pelvic tilt offset. Therefore, altering the pelvis position throughout the drive phase of the serve, via coaching (in combination with motor control interventions ), focused on adjustments to the timing of knee joint extension and/or the alignment of players’ feet relative to their centre of mass, may reduce the potentially injurious lower lumbar movements/forces. Finally, prescriptive stability exercise programs may not be helpful for the prevention of LBP in this population (4), as these may add to increased joint stiffness (17) and compressive forces already present in this group. Whether or not mobilization or neuromuscular relaxation/control strategies to increase lumbar mobility is a useful management strategy is yet to be determined. Intervention and prospective studies including lower lumbar mobility and lumbar/pelvic/lower limb serving kinematics are required to confirm these hypotheses.
This study was limited to a relatively small population of elite young male players from one country and should be replicated in other groups. Further, the study was cross-sectional in nature; therefore, the relationships identified should be examined in prospective studies to provide evidence on the direction of causality.
This project was undertaken with the support of the Curtin University Linkage Grants Scheme, a Tennis Australia’s Sport Science and Medicine Research Grant, and the Australian Institute of Sport.
The authors have no disclosures or conflicts of interest.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Alyas F, Turner M, Connell D. MRI findings in the lumbar spines of asymptomatic, adolescent, elite tennis players. Br J Sports Med
. 2007; 41 (11): 836–41.
2. Besier TF, Lloyd DG, Ackland TR. Muscle activation strategies at the knee during running and cutting maneuvers. Med Sci Sports Exerc
. 2003; 35 (1): 119–27.
3. Burnett A, O’Sullivan P, Ankarberg L, et al. Lower lumbar spine axial rotation is reduced in end-range sagittal postures when compared to a neutral spine posture. Man Ther
. 2008; 13 (4): 300–6.
4. Cairns MC, Foster NE, Wright C. Randomized controlled trial of specific spinal stabilization exercises and conventional physiotherapy for recurrent low back pain. Spine
. 2006; 31 (19): E670–81.
5. Campbell A, Straker L, O’Sullivan P, Elliott B, Reid M. Lumbar loading in the elite adolescent tennis serve: a link to low back pain. Med Sci Sports Exerc
. 2013; 45 (8): 1562–8.
6. Crewe H, Campbell A, Elliott B, Alderson J. Lumbo-pelvic biomechanics and quadratuslumborum asymmetry in cricket fast bowlers. Med Sci Sports Exerc
. 2012; 45 (4): 778–83.
7. Dankaerts W, O’Sullivan P, Burnett A, Straker L, Davey P, Gupta R. Discriminating healthy controls and two clinical subgroups of nonspecific chronic low back pain patients using trunk muscle activation and lumbosacral kinematics of postures and movements: a statistical classification model. Spine
. 2009; 34 (15): 1610–8.
8. Dankaerts W, O’Sullivan PB, Burnett AF, Straker LM. The use of a mechanism-based classification system to evaluate and direct management of a patient with non-specific chronic low back pain and motor control impairment–a case report. Man Ther
. 2007; 12 (2): 181–91.
9. Fleisig G, Nicholls R, Elliott B, Escamilla R. Kinematics used by world class tennis players to produce high-velocity serves. Sports Biomech
. 2003; 2 (1): 51–64.
10. Foster D, John D, Elliott B, Ackland T, Fitch K. Back injuries to fast bowlers in cricket: a prospective study. Br J Sports Med
. 1989; 23 (3): 150–4.
11. Hjelm N, Werner S, Renstrom P. Injury profile in junior tennis players: a prospective two year study. Knee Surg Sports Traumatol Arthrosc
. 2010; 18 (6): 845–50.
12. Hodges P, van den Hoorn W, Dawson A, Cholewicki J. Changes in the mechanical properties of the trunk in low back pain may be associated with recurrence. J Biomech
. 2009; 42 (1): 61–6.
13. Kibler BW. The 4000 Watt tennis player: power development for tennis. Med Sci Tennis
. 2009; 14 (1): 5–8.
14. Kibler WB, Chandler TJ. Range of motion in junior tennis players participating in an injury risk modification program. J Sci Med Sport
. 2003; 6 (1): 51–62.
15. Kibler WB, Safran MR. Musculoskeletal injuries in the young tennis player. Clin Sports Med
. 2000; 19 (4): 781–92.
16. Markolf KL. Deformation of the thoracolumbar intervertebral joints in response to external loads: a biomechanical study using autopsy material. J Bone Joint Surg Am
. 1972; 54 (3): 511–33.
17. McGill SM. Low back exercises: evidence for improving exercise regimens. Phys Ther
. 1998; 78 (7): 754–65.
18. McGregor AH, Hukins DW. Lower limb involvement in spinal function and low back pain. J Back Musculoskelet Rehabil
. 2009; 22 (4): 219–22.
19. Mitchell T, O’Sullivan PB, Burnett AF, Straker L, Smith A. Regional differences in lumbar spinal posture and the influence of low back pain. BMC Musculoskelet Disord
. 2008; 9: 152.
20. O’Sullivan P. Diagnosis and classification of chronic low back pain disorders: maladaptive movement and motor control impairments as underlying mechanism. Man Ther
. 2005; 10: 242–55.
21. Panjabi MM. The stabilizing system of the spine. Part I. Function, dysfunction, adaptation, and enhancement. J Spinal Disord
. 1992; 5 (4): 383–9.
22. Panjabi MM. The stabilizing system of the spine. Part II. Neutral zone and instability hypothesis. J Spinal Disord
. 1992; 5 (4): 390–7.
23. Portus M, Mason BR, Elliott BC, Pfitzner MC, Done RP. Technique factors related to ball release speed and trunk injuries in high performance cricket fast bowlers. Sports Biomech
. 2004; 3 (2): 263–84.
24. Reid M, Elliott B, Alderson J. Lower-limb coordination and shoulder joint mechanics in the tennis serve. Med Sci Sports Exerc
. 2008; 40 (2): 308–15.
25. Reid M, Whiteside D, Elliott B. Serving to different locations: set-up, toss, and racket kinematics of the professional tennis serve. Sports Biomech
. 2011; 10 (4): 407–14.
26. Reid MM, Campbell AC, Elliott BC. Comparison of endpoint data treatment methods for estimation of kinematics and kinetics near impact during the tennis serve. J Appl Biomech
. 2012; 28 (1): 93–8.
27. van Dieen JH, Cholewicki J, Radebold A. Trunk muscle recruitment patterns in patients with low back pain enhance the stability of the lumbar spine. Spine (Phila Pa 1976)
. 2003; 28 (8): 834–41.
28. van Dieen JH, Selen LP, Cholewicki J. Trunk muscle activation in low-back pain patients, an analysis of the literature. J Electromyogr Kinesiol
. 2003; 13 (4): 333–51.
29. Wade M, Campbell A, Smith A, Norcott J, O’Sullivan P. Investigation of spinal posture signatures and ground reaction forces during landing in elite female gymnasts. J Appl Biomech
. 2012; 28 (6): 677–86.
30. Woltering HJ. A fortran package for generalized, cross-validatory spline smoothing and differentiation. Adv Eng Softw
. 1986; 8 (2): 104–13.
31. Wu G, Siegler S, Allard P, et al. ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion—part I: ankle, hip and spine. J Biomech
. 2002; 35: 543–8.
32. Zusman M. Associative memory for movement-evoked chronic back pain and its extinction with musculoskeletal physiotherapy. Phys Ther Rev
. 2008; 13 (1): 57–69.
Keywords:© 2014 American College of Sports Medicine
BIOMECHANICS; INJURY PREVENTION; BACK PAIN; TENNIS