Thigh muscle strength imbalance has been implicated as predictive of some common lower limb musculoskeletal injuries in field and court sports (1,8,9,18,19,22,24–26). Research in this area has examined strength imbalances on isokinetic devices measuring a range of variables including hamstring and quadriceps torques, conventional hamstring-quadriceps ratio (H con:Q con), functional hamstring-quadriceps ratio (H ecc:Q con), bilateral ratios, and stronger-weaker ratios (12,16). The reliability of these measurements at various velocities has been established, and some have been shown to be more reliable than others. For instance, absolute measures have been shown to be more reliable than strength balance ratios (12,16,28). For absolute measures, the fewer the repetitions, the better the reproducibility (12,28); concentric actions have been shown to have greater reliability than eccentric actions (12,28); measurements taken at lower velocities are typically more reliable than are measures at higher velocity (12,16,17); and less variability has been observed with extensor movements compared with flexor movements (12,28). For strength ratios, H ecc:Q con is reported to be more reliable than are others, possibly because H ecc:Q con more accurately reflects the dynamic function of hamstring and quadriceps muscle groups and consequently better describes dynamic muscular stabilization of the knee (16).
There is compelling evidence to suggest a relationship between muscle imbalance and lower limb soft tissue injury (9,24), and studies have indicated that effective activation of the eccentric component of the hamstrings during active knee extension reduces loading on the anterior cruciate ligament (ACL) (1,18). Furthermore, training studies have shown that strength balance ratios can be improved and that improvements may reduce the incidence of lower limb musculoskeletal injury (2,14). Despite this evidence, doubt over the value of the hamstring-quadriceps strength balance ratios as a screening tool for injury risk remains. This may, in part, be because of a perceived poor relationship between isokinetic strength and muscular power (19,21) and isokinetic strength and sprinting performance (20). Other limitations may include the movement velocity used in available studies, which does not represent the limb movement velocity during real-world movements such as sprinting or the influence of hip joint position. It may also have not been helped by the inconsistency in study methodology and outcomes (1,8,9,23,28).
It is hardly surprising therefore that current data exploring the relationship between hamstring-quadriceps balance ratios and injury are conflicting. For example, Orchard et al. (24) reported a significant relationship between the H con:Q con ratio and hamstring injury, but Bennell et al. (4) found no relationship between the same outcome variables. More recently, Croisier et al. (9) reported a strong correlation between the H ecc:Q con ratio, determined from the eccentric hamstring torque at a slow velocity (0.53 rad·s−1) and concentric quadriceps torque at a fast velocity (4.19 rad·s−1), and hamstring injury. These conflicting data may be largely because of Bennell et al. (4) using H con:Q con. Croisier et al.'s (9) work demonstrated that the H con:Q con would not have detected approximately 30% of hamstring injuries in their study.
One major consideration that has been ignored in previous studies of either H con:Q con or H ecc:Q con is the influence of the hip joint position (1,20,24). Studies that have investigated the relationship between isokinetic test performance and lower limb musculoskeletal injury have typically reported data obtained from participants tested in a seated position. However, rarely are field and court sport athletes active with those kinematics (e.g., the hip flexed at 90°). Most lower limb injuries occur while athletes engage in some running activity, specifically, at foot plant (1,4,7,9,18,23). For overground running, trunk angle is reported to typically be approximately 10° to the vertical with foot plant occurring directly inferior to the torso (Figure 1) (29). Thus, when the hip and knee joints are nearer full extension, dynamic knee joint stability is most important. Consequently, it could be argued that isokinetic screening where the hip angle is more similar to when executing real-world sporting tasks, using an eccentric hamstring strength testing protocol would be more ecologically valid than using other traditional methods.
Altering the hip angle for lower limb isokinetic screening might have an effect on hamstring and quadriceps torques and subsequent knee joint strength ratios. At the very least, the stretch-tension relationship of the hamstrings and quadriceps muscle groups will likely differ (20). Therefore, the relative contribution of the active contractile components of the muscle to overall force production would change. This theory is supported by work that has examined the effect of hip position on knee torque production (3,6,15,20) and changes in neuromuscular activation (determined from electromyography) throughout the range of motion (ROM) (17). However, studies that have compared the effect of hip position on isokinetic test performance are limited to only determining whether a significant difference between positions exists (3,6,15,20). No studies have explored the level of agreement between peak torque measures from supine and seated positions using Bland and Altman's limits of agreement (LOAs) (3,5,6,15,20). If H ecc:Q con is to be used as a screening tool for injury risk, it is important to determine the level of agreement of values obtained when the hip joint is placed in different positions. If there is good agreement between positions, then strength balance ratios would not change, and ratios calculated from each position would be equally able to predict musculoskeletal injury. In addition to being limited by statistical constraints, published research that has investigated the effect of hip position on knee joint strength ratios is limited to examining the H con:Q con ratio only (15).
Consistent application of a screening method that measures eccentric hamstring strength in a hamstring-quadriceps ratio is necessary because some of the most severe and costly injuries in sport typically occur during active extension of the knee and during the terminal swing phase during running or sprinting (9,23,24,26). Understanding the effect of hip angle on hamstring and quadriceps concentric and eccentric torques and knee joint strength ratios, and applying such knowledge, might enhance current screening methods and subsequently lead to the development of a standard, more ecologically valid, isokinetic protocol. Information obtained from such screening methods may enable sports practitioners to more effectively identify athletes at a greater risk of sustaining lower limb musculoskeletal injury and allow them to alter training practices to reduce injury risk or to establish progress from rehabilitation. Therefore, the aims of this study were to compare isokinetic strength measurements recorded in a near supine position where kinematics were more similar to what would be observed while executing real-world sporting tasks (i.e., hip flexion 10° to the vertical) to seated measurements to determine the effect of hip position on H con:Q con and H ecc:Q con.
Experimental Approach to the Problem
This was a cross-sectional, repeated measures study. The participants attended the laboratory on 3 occasions, the first being for familiarization, and the other 2 were test sessions, the order of which was randomized for seated or supine position. There were between 7 and 14 days between sessions.
Eleven academy players from an English Premiership Rugby Union Club (characteristics mean ± σ, age 19.3 ± 0.8 years, body mass 92.8 ± 12.6 kg, stature 182.22 ± 8.07 cm) volunteered to participate in this study. All the participants completed the testing in the 3 weeks immediately before the commencement of preseason games. All the players were free from injury or illness. Written informed consent was obtained from all the participants, and a health questionnaire screening was conducted. The University Research Ethics Committee approved the study.
Stature and body mass were measured using a stadiometer (Holtain, Crymych, Dyfed, United Kingdom) and scales (Cranlea, Birmingham, United Kingdom). A warm-up before testing was performed on a Monark cycle ergometer (Monark 814E, Varberg, Sweden). Isokinetic measurements were made on a Biodex System 3 (Shirley, NY, USA). All statistical analyses were performed using SPSS for Windows (V16.0, SPSS Inc., Chicago, IL, USA).
Warm-Up and Dynamometer Positioning
For 48 hours before testing, all the participants refrained from intense exercise, especially eccentric exercise to reduce the likelihood of delayed onset muscle soreness affecting the results. All the participants were asked to remain adequately hydrated before testing but refrain from drinking caffeine 12 hours before testing. Food was not consumed 2 hours before testing. All the tests involved a standardized procedure, including a 3-minute warm-up on a cycle ergometer at a self-regulated moderate intensity.
In the seated test, the participants were placed in a seated position with the backrest positioned at 1.4 rad flexion. The axis of rotation of the dynamometer was aligned with the lateral epicondyle of the dominant knee, and the cuff was placed approximately 2 cm superior to the medial malleolus. Straps were tightened around the chest, pelvis, and thigh for stabilization. The ROM was set using the voluntary knee extension position that the participant deemed to be comfortably straight but not hyperextended 0 to 1.31 rad knee flexion. The ROM was limited in this way to more easily enable the extended or flexed knee to achieve the necessary preload in the eccentric test. The same ROM was used for the other leg. A hard cushion was used so that the length of the acceleration and deceleration phase was shortened. Once positioned, the gravity correction procedure involved the participants relaxing their leg so that it could be weighed during passive knee flexion, in accordance with the manufacturer's recommendations.
For the other test, the participants were placed lying supine with the backrest positioned at 0.2 rad flexion. All the other procedures were the same as that described for the seated test. Figure 1 describes the rationale behind the supine angle and shows the 2 testing positions.
For each velocity and mode of muscle action, the participants were permitted 4 familiarization repetitions of increasing effort with 30 seconds of rest before the test and 90 seconds between the test and the next set of familiarization repetitions. During the test, the participants were instructed to push and pull or resist the attachment as hard and as fast as possible. Three continuous maximal efforts at 1.04 rad·s−1 and 4 at 3.14 rad·s−1 were performed with concentric tests taking place before eccentric tests. Knee extensors always acted first. Verbal encouragement was given by the same experimenter but no visual feedback was given. Both knees were tested on the same day, but the starting leg was randomized. The order of seated and supine testing was also randomized between the participants.
Hip flexion angle and testing velocity were the independent variables. Peak torques for concentric and eccentric muscle actions for the hamstrings and quadriceps muscle groups and knee joint strength ratios were the dependent variables.
The highest gravity corrected peak torque on the windowed and filtered output was rounded to the nearest 1 N·m and recorded for further analysis. To reduce the effects of acceleration and deceleration of the lever arm on torque output, only peak torque data obtained from a period of constant velocity (within a 5% range of the preset angular velocity) were used for analysis. Descriptive statistics were presented as mean ± σ for all peak torque values and torque ratios. Both H con:Q con and H ecc:Q con were calculated using peak torque data. A repeated measures analysis of variance (ANOVA) was performed on peak torque data. There were 4 within-subject factors: position (seated or supine), agonist (quadriceps or hamstrings), muscle action (concentric or eccentric), and velocity (1.05 or 3.14 rad·s−1). Where significant interaction or main effects were found, paired t-tests with Bonferroni adjustment were used to assess the differences between pairs. The same analysis was performed on H con:Q con and H ecc:Q con but with position and velocity as within-subject factors. Pearson correlations were calculated between seated and supine variables. Ninety-five percent ratio LOAs (5) based on log transformed data and antilogged to give a dimensionless ratio, which represents random error, were calculated to determine the extent of agreement between seated and supine variables. Limits of agreements were only calculated where there was no significant difference between seated and supine variables. The alpha level was set at p ≤ 0.05.
Mean values of peak torque, Pearson correlations, and 95% ratio LOA are given in Table 1.
The repeated measures ANOVA demonstrated significant main effects of position (seated greater than supine, p = 0.014), agonist (extensor greater than flexor, p < 0.001), muscle action (eccentric greater than concentric, p = 0.002), and velocity (slower greater than fast, p < 0.001) were also identified. There was a significant position × muscle action interaction (p < 0.05) for peak torque (see Figure 2).
The interaction was because of the eccentric peak torque being reduced more in the supine position compared with concentric peak torque. Paired t-tests revealed a significantly lower extensor concentric peak torque at 3.14 rad·s−1, extensor eccentric peak torque at 1.05 rad·s−1, and flexor eccentric peak torque at 3.14 rad·s−1 in the supine position (Table 1). Pearson correlations between seated and supine peak torques varied from low to high. Where no significant difference between seated and supine peak torque existed, 95% ratio LOAs were calculated and varied from ×/÷1.38 to 1.53. That is, seated and supine peak torque measurements will differ because of random error by between 38 and 53% on either side of the systematic bias, which ranged from 6 to 21%.
Mean values of H con:Q con and H ecc:Q con, Pearson Correlations, and 95% ratio LOAs are shown in Table 2.
No other significant interactions were observed, but the Repeated Measures ANOVA (Table 3) revealed a significant main effect of velocity (p < 0.05) on H con:Q con and H ecc:Q con because of a higher ratio at the faster velocity. There was also a significant (p < 0.05) main effect of position for H ecc:Q con only. Paired t-tests revealed the seated H ecc:Q con was significantly greater than the supine equivalent at the faster velocity only (Table 2). However, it was for this ratio that there was a significant Pearson correlation (p < 0.05) between seated and supine positions. All other correlations were low. Larger differences between seated and supine positions were observed in the H ecc:Q con compared with the H con:Q con with the mean seated H:Q being greater than that for supine. Where there was no significant difference between seated and supine H:Q, 95% ratio LOAs were calculated and varied from ×/÷1.37 to 1.51 on either side of the systematic bias, which ranged from 9 to 14% (Table 2).
The first aim of this study was to compare knee flexion and extension isokinetic peak torque measured for supine compared with that for the seated position. A significant position × muscle action interaction effect was found with a greater concentric torque recorded in a seated position compared with eccentric torque in a supine position. Subsequently, the significant main effects for position (seated torque greater than supine) and muscle action (eccentric greater than concentric) are in agreement with the findings of the existing literature (3,6,15,20). We found that for 3 of the 8 peak torque variables, the mean peak torque was significantly greater in the seated position compared with that in the supine position; for the other 5 measures, the agreement was poor, that is, the random error limits were between 37 and 53%, and there was a large systematic bias ranging between 6 and 21% (Table 1). Furthermore, in most instances, correlations were only weak to moderate. Therefore, it can be argued that the results obtained in a seated position would typically be significantly different and unrelated to testing in the near supine position.
Both concentric and eccentric peak torques were negatively affected by testing in the supine position. However, the magnitude of this effect was greater for eccentric actions (Figure 2). This is not surprising because supine peak torques were dissimilar and unrelated to seated peak torques as indicated by some significant differences and poor agreements and correlations.
This study, similar to others (3,6,15,20), has shown that hip angle influences both concentric and eccentric peak torques. Based on the results from this study and others (3,6,15,17,20), it can be hypothesized that the hip angle influences the stretch-tension relationship of the muscle, the relative contribution of active contractile components of the muscle, and neuromuscular control, which ultimately effects a number of isokinetic peak torque indices. For example, it could be argued that when extending the knee with a greater hip-thigh angle, the neural activation of the hamstrings differs from that when seated because of less tension applied by the series elastic and parallel elastic components of posterior chain muscles. Further research to support this argument is necessary. Repeating this study with a larger sample while concurrently measuring muscle activity using electromyography would be a reasonable approach.
In agreement with the results of other studies, we have found a significant main effect for velocity on both concentric and eccentric torque production, with greater torque found at the slower velocity (12,15). However, it is important that for knee stability during faster velocity movements, the eccentric hamstring torque is relatively unaffected by velocity to increase the H ecc:Q con to produce less strain on the ACL. Irrespective of hip positioning, we have found that concentric quadriceps decreases by around 20% with increasing velocity, but comparable eccentric hamstring torque only decreased by 3%.
The second aim of this study was to determine if there was an influence of the hip angle on knee joint strength ratios. This is important to determine if strength ratios are to be used as a screening tool to explore the possible risk of an individual to injury. Because the hip is rarely fixed at 90° during most functional movements, the assessment of the ratio in a seated position provides little ecological validity. Determination of strength ratios in a prone or supine position where the hip is fixed at a position that more closely reflects running (10° of hip flexion) is more valid, especially because it replicates more closely the length-stretch relationship. It is important when testing in a supine position to correct for gravity, as we have done in this study, because the gravitational influence on torque production will be different from that for upright running. Unlike previous work, a main effect for H con:Q con was not found in this study (15). However, a main effect of position on H ecc:Q con was observed. To the knowledge of the researchers of this study, this is the first study which has examined the effect of hip angle on H ecc:Q con. The nonsignificant effect of position on H con:Q con is not surprising because its calculation requires division of one concentric peak torque by another (14). Assuming that hamstring and quadriceps concentric peak torques in the 2 positions differed by the same amount, the same ratio was expected for H con:Q con, whereas H ecc:Q con calculation requires the division of an eccentric action by a concentric action (14) and because eccentric actions were more negatively affected by position, a smaller H ecc:Q con from testing in the supine position was expected because the numerator in the equation was disproportionately smaller.
Unlike the main effect of position for which an effect was observed for H ecc:Q con only, a main effect of velocity was observed for H ecc:Q con and H con:Q con. This can be explained by the main effect of velocity on the absolute values from which the ratios are calculated. However, these results must be interpreted with caution because torque reliability at higher velocities becomes questionable (9,12,28).
As noted previously, isokinetic measurement of knee joint strength balance can be used as a screening tool to predict lower limb musculoskeletal injury (1,8,9,18,22,24–26). However, evidence to support the relationship between muscular imbalances and lower limb musculoskeletal injury is inconsistent (1,4,9,13,24). Thus, the development of a standard, ecologically valid testing protocol is necessary (9,23–26). Using an eccentric protocol, Croisier et al. (9) revealed a strong relationship between strength imbalance and hamstring strain. However, their mixed H ecc:Q con still did not detect approximately 5% of injuries, and despite having a large sample, their alpha level was set at p < 0.05. Therefore, their protocol, although promising, may have ‘missed’ a considerable number of injuries. This may be explained by the fact that the protocol used by Croisier et al. (9) tested participants in a seated position, given that the present study has shown that hip flexion angle affects isokinetic test performance considerably. This begs the question—Because hip angle affects concentric and eccentric peak torque, and this has a carryover effect to H ecc:Q con, would an H ecc:Q con calculated from peak torques measured with a hip angle which more closely reflects that which is observed while executing real-world sporting tasks better predict musculoskeletal injury?
Compelling evidence showing a relationship between knee joint strength ratios determined by the use of isokinetic dynamometers and lower limb musculoskeletal injury exists. Furthermore, training studies have shown that knee joint strength ratios can be improved and, consequently, injury risk may be reduced. Despite this evidence, some reluctance by sports practitioners to test knee joint strength ratios on isokinetic dynamometers remains. This may be because of perceptions of a lack of relationship between isokinetic test performance and other physical performance qualities. It may also be related to inconsistencies in testing protocols and outcomes. Thus, we argued that the development of a standard ecologically valid testing protocol be developed. Evidence leans toward testing protocols that measure hamstring strength eccentrically being better able to predict injury. However, in studies that have presented this evidence, a considerable number of injuries were still not predicted. We highlighted that an oversight of much of the research to date is the effect of hip position on isokinetic test performance. In fact, it has been argued in this article that the ecological validity of isokinetic testing protocols for knee joint strength ratios is typically questionable because they typically test athletes in seated positions (i.e., hip angle of 90°). Most functional tasks in field and court sports, rugby included, are executed with far less hip flexion (i.e., hip angle of ∼10°). This study showed that hip position has a significant effect on isokinetic peak torque and that agreement between seated and supine measurements was poor. Furthermore, the effect of hip position on peak torques carried over to affect functional knee joint strength ratio. Thus, an isokinetic testing protocol that considers eccentric hamstring strength where measurements are recorded with a hip flexion angle nearer 10° is likely to be most ecologically valid. Using such a protocol, strength imbalances can be determined and lower limb musculoskeletal injury may be predicted. By adopting screening methods such as this, sports practitioners can affect training to reduce injury risk and therefore enhance performance.
No financial assistance was awarded for this project. The authors wish to thank the participants for volunteering their time.
1. Alentorn-Geli E, Myer GD, Silvers HJ, Samitier G, Romero D, Lazaro-Haro C, Cugat R. Prevention of non-contact anterior cruciate ligament injuries in soccer playersPart 1: Mechanisms of injury
and underlying risk factors. Knee Surg Sports Traumatol Arthrosc 17: 705–729, 2009.
2. Askling C, Karlsson J, Thorstensson A. Hamstring injury
occurrence in elite soccer players after preseason strength training with eccentric overload. Scand J Med Sci Sports 13: 244–250, 2003.
3. Barr AE, Duncan PW. Influence of position on knee flexor peak torque. J Orthop Sports Phys Ther 9: 279–283, 1988.
4. Bennell K, Wajswelner H, Lew P, Schall-Riaucour A, Leslie S, Plant D, Cirone J. Isokinetic
strength testing does not predict hamstring injury
in Australian Rules footballers. Br J Sports Med 32: 309–314, 1998.
5. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307–310, 1986.
6. Bohannon RW, Gajdosik RL, LeVeau BF. Isokinetic
knee flexion and extension torque in the upright sitting and semireclined sitting positions. Phys Ther 66: 1083–1086, 1986.
7. Cochrane JL, Lloyd DG, Buttfield A, Seward H, McGivern J. Characteristics of anterior cruciate ligament injuries in Australian football. J Sci Med Sport 10: 96–104, 2007.
8. Croisier JL. Factors associated with recurrent hamstring injuries. Sports Med 34: 681–695, 2004.
9. Croisier JL, Ganteaume S, Binet J, Genty M, Ferret JM. Strength imbalances and prevention of hamstring injury
in professional soccer players: A prospective study. Am J Sports Med 36: 1469–1475, 2008.
10. Duthie G, Pyne D, Hooper S. Time motion analysis of 2001 and 2002 super 12 rugby. J Sports Sci 23: 523–530, 2005.
11. Duthie GM, Pyne DB, Marsh DJ, Hooper SL. Sprint patterns in rugby union players during competition. J Strength Cond Res 20: 208–214, 2006.
12. Gleeson NP, Mercer TH. Reproducibility of isokinetic
leg strength and endurance characteristics of adult men and women. Eur J Appl Physiol Occup Physiol 65: 221–228, 1992.
13. Grace TG, Sweetser ER, Nelson MA, Ydens LR, Skipper BJ. Isokinetic
muscle imbalance and knee-joint injuriesA prospective blind study. J Bone Joint Surg Am 66: 734–740, 1984.
14. Holcomb WR, Rubley MD, Lee HJ, Guadagnoli MA. Effect of hamstring-emphasized resistance training on hamstring:quadriceps strength ratios. J Strength Cond Res 21: 41–47, 2007.
15. Hopkins J, Sitler M, Ryan J. The effects of hip position and angular velocity on quadriceps and hamstring eccentric peak torque and ham/quad ration. Isokinet Exerc Sci 3: 27–33, 1993.
16. Impellizzeri FM, Bizzini M, Rampinini E, Cereda F, Maffiuletti NA. Reliability of isokinetic
strength imbalance ratios measured using the Cybex NORM dynamometer. Clin Physiol Funct Imaging 28: 113–119, 2008.
17. Kellis E, Katis A. Quantification of functional knee flexor to extensor moment ratio using isokinetics and electromyography. J Athl Train 42: 477–485, 2007.
18. Kelly AK. Anterior cruciate ligament injury
prevention. Curr Sports Med Rep 7: 255–262, 2008.
19. Lehance C, Binet J, Bury T, Croisier JL. Muscular strength, functional performances and injury
risk in professional and junior elite soccer players. Scand J Med Sci Sports 19: 243–251, 2009.
20. Manou V, Saraslanidis P, Zafeiridis A, Kellis S. Sitting vs. prone isokinetic
strength in elite male and female sprinters: Relationship with sprinting performance. J Hum Mov Stud 45: 273–290, 2003.
21. Morriss CJ, Tolfrey K, Coppack RJ. Effects of short-term isokinetic
training on standing long-jump performance in untrained men. J Strength Cond Res 15: 498–502, 2001.
22. MSFC B. Clincal practice guides for muscular injuries, epidemiology, diagnosis, treatment and prevention. Apunts Med Esport 164: 179–203, 2009.
23. Orchard J. Biomechanics of muscle injury
strain. New Zeal J Sports Med 30: 92–98, 2003.
24. Orchard J, Marsden J, Lord S, Garlick D. Preseason hamstring muscle weakness associated with hamstring muscle injury
in Australian footballers. Am J Sports Med 25: 81–85, 1997.
25. Orchard J, Seward H, McGivern J, Hood S. Intrinsic and extrinsic risk factors for anterior cruciate ligament injury
in Australian footballers. Am J Sports Med 29: 196–200, 2001.
26. Orchard JW. Intrinsic and extrinsic risk factors for muscle strains in Australian football. Am J Sports Med 29: 300–303, 2001.
27. Roberts SP, Trewartha G, Higgitt RJ, El-Abd J, Stokes KA. The physical demands of elite English rugby union. J Sports Sci 26: 825–833, 2008.
28. Sauret J, De Ste Croix M, Deighan M, Iga J, James D. Reproducibility of an isokinetic
eccentric muscle endurance task. Eur J Sport Sci 9: 311–319, 2009.
29. Williams K. Biomechanics of running. Exerc Sport Sci Rev 13: 389–431, 1985.