Muscle Activation Strategies at the Knee during Running and Cutting Maneuvers

THOR F. BESIER, DAVID G. LLOYD, and TIMOTHY R. ACKLAND School of Human Movement & Exercise Science, The University of Western Australia, Perth, AUSTRALIA

ABSTRACT BESIER, T. F., D. G. LLOYD, and T. R. ACKLAND. Muscle Activation Strategies at the Knee during Running and Cutting Maneuvers. Med. Sci. Sports Exerc., Vol. 35, No. 1, pp. 119 127, 2003. Purpose: The purpose of this article was to investigate the activation patterns of muscles surrounding the knee during preplanned (PP) and unanticipated (UN) running and cutting tasks, with respect to the external moments applied to the joint. It was hypothesized that activation strategies during PP tasks would correspond to the magnitude and direction of the external loads applied to the knee joint, and the muscle activation patterns would differ between PP and UN tasks. Methods: Eleven healthy male subjects performed a series of running and cutting tasks under PP and UN conditions. Activation from 10 knee muscles were determined using full-wave rectified, filtered, and normalized EMG calculated during a precontact phase and two epochs across the stance phase. Knee joint flexor and extensor muscle group ratios indicated the level of co-contraction. Individual muscles were also grouped into medial/lateral and internal/external rotation muscle groups, based upon their ability to counter externally applied varus/valgus and internal/external rotation joint loads, respectively. Results: Selective activation of medial/lateral and internal/external rotation muscles and co-contraction of flexors and extensors were used to stabilize the joint under PP conditions, whereas generalized co-contraction strategies were employed during the UN condition. Net muscle activation during the UN sidestepping tasks increased by 10 20%, compared with an approximately 100% increase in applied varus/valgus and internal/ external rotation joint moments. Conclusion: In PP conditions, activation patterns appear to be selected to support the external loads experienced at the knee, e.g., medial muscles activated to resist applied valgus moments. Under UN conditions, there was no selective activation of muscles to counter the external knee load, with generalized co-contraction being the activation pattern adopted. These findings have implications for the etiology of noncontact knee ligament injuries. Key Words: EMG, KNEE LIGAMENT INJURY, ACTIVATION PATTERNS, RUNNING

oncontact knee ligament injuries are common to many sports, yet the mechanism behind these injuries is not well understood. From a purely mechanistic standpoint, ligament failure occurs when the applied load exceeds its strength. Ligament strength depends on many factors such as hormone levels, loading history, and prior injury, among others. The load applied to the ligament depends on the external load applied to the knee joint, the kinematics and geometry of the articulating surfaces when these external forces are applied, and the activation of muscles surrounding the joint. Therefore, one way to reduce the risk of injury is to assess the factors that affect the load applied to the ligament. In an effort to characterize the external loads placed on the knee during typical sporting maneuvers, the three-dimensional moments applied to the joint during running,

Address for correspondence: David Lloyd, Ph.D., School of Human Movement & Exercise Science, The University of Western Australia, 35 Stirling Highway, Crawley, Western Australia 6009; E-mail: dlloyd@cyllene.uwa.edu.au. Submitted for publication November 2001. Accepted for publication September 2002. 0195-9131/03/3501-0119/$3.00/0 MEDICINE & SCIENCE IN SPORTS & EXERCISE Copyright 2003 by the American College of Sports Medicine DOI: 10.1249/01.MSS.0000043608.79537.AB

sidestep cutting, and crossover-cutting tasks have been previously measured (4). Table 1 provides a summary of the relative loading patterns at the knee during the performance of these maneuvers throughout the stance phase (3,4). Based on cadaveric studies, combined applied moments of flexion, valgus, and internal rotation are known to place the greatest strain on the anterior cruciate ligament or ACL (17), and these occur during sidestepping tasks (4). Combined applied flexion, varus, and external rotation moments were found to occur during the crossover cut, which may place strain on the lateral collateral ligament (4). The magnitude of the applied moments changed with the different maneuvers (4). As the sidestep angle increased from 30 to 60, the applied valgus and internal rotation moments increased. Similarly, the applied varus and external rotation moments increased going from the run to the crossover cut. However, there was little difference in the applied flexion moments during the two sidestepping tasks, crossover cut and the run. The effect of anticipation on moments applied to the knee joint during running and cutting tasks has also been investigated (3). In unanticipated (UN) cutting tasks, the subject was required to perform rapid and unexpected maneuvers, resulting in increased varus/valgus (VV) and internal/external rotation (IE) moments at the knee. The increased moments were as much as 100% greater than cutting maneuvers that were preplanned (PP) (3) (Table 1). These UN conditions are more likely to resemble the movements that occur during a game situation compared with PP conditions 119

TABLE 1. Summary of applied moments to the knee during sidestepping to 60 and 30 (S60 and S30, respectively), running (RUN), and crossover cutting to 30 (XOV) (Besier et al. 3,4) during the weight acceptance (WA) and peak push off (PPO) phases; applied flexion moments have not been included as they were similar between tasks and anticipated/unanticipated conditions. External Joint Moments (Nmkg1) Maneuver Type S60 S30 RUN XOV PreplannedUnanticipated Preplanned Unanticipated Preplanned Unanticipated Preplanned Unanticipated Preplanned Unanticipated Varus WA PPO WA 0.3 0.5 0.03* 0.3* 0.2 0.25 0.45 0.8 0.6 0.75 1.2 1.8 Valgus PPO 0.3 0.45 0.3* 0.45* 0.07 0.07 0.11 0.2 0.08 0.09 0.38 0.55 External Rotation WA PPO Internal Rotation WA 0.13 0.21 0.08 1.8 PPO 0.3 0.31 0.25 0.33

* Data taken from valgus subject group; refer to Besier et al. (3,4) for details. Significant difference between preplanned and unanticipated condition (P 0.05).

and demonstrate the increased risk to knee joint ligaments if there is limited time for the person to make appropriate postural adjustments (3). Muscles surrounding the knee have the ability to support the large external loads applied to the joint and reduce the potential for ligament loading, owing to their anatomical moment arms (1,14,16). Table 2 lists the major muscles surrounding the knee joint and their respective moment arms for producing flexion/extension (FE), VV, and IE moments. When destabilizing forces are anticipated, the CNS is capable of adjusting muscle activation patterns to oppose these forces, supporting the notion that anticipatory postural adjustments are planned in detail (2). During a dynamic landing task, for example, the hamstring muscles are activated before landing, which is believed to counter the anterior translation of the tibia with respect to the femur that occur just after landing and protect the ACL (10). Cowling and Steele (10) also found differences in hamstring activation timing between genders, which they postulate to be responsible for the variations in ACL injury rates among men and women. The preprogrammed nature of these postural adjustments suggest that there may be insufficient time for the CNS to plan appropriate activation strategies to stabilize the greatly increased joint loads during UN cutting tasks (3). Indeed, lack of appropriate postural adjustments is hypothesized to be responsible for the large increase in externally applied joint loads during UN sidestepping tasks (3).

Given the over-determined nature of the knee joint (i.e., greater number of muscles than degrees of freedom), the CNS may adopt a number of different neural strategies to counter the external loads during dynamic cutting tasks. However, two generalized strategies have been suggested (13,16). The first strategy involves selected activation of muscles with moment arms that are best able to counter the external load. Examples of this would include increased activation of gracilis to counter an external valgus moment, or co-activation of the medial hamstrings and medial quadriceps to oppose external valgus loading. The second strategy is generalized co-contraction (13,16), involving coactivation of hamstring and quadriceps muscles without any selectivity of specific muscles. Selected activation and generalized co-contraction have been shown to stabilize the knee joint during isometric VV tasks (1,6,15,16). It is hypothesized that different neural strategies will be employed during PP and UN sidestepping tasks, based upon the anticipation, or lack of anticipation, of the forthcoming movement. Joint mechanics may also play a role in the selection of muscle activation patterns to support external loads at the knee. The condylar joint, for example, is well designed to counter external varus and valgus loads as the effective axis of rotation for muscle moment arms can shift between the medial and lateral condyle (Fig. 1). Owing to the quadriceps insertion to the midline of the tibia, these muscles can therefore counter both varus and valgus loads at the knee

TABLE 2. Moments produced by major muscles surrounding the knee joint at knee flexion angles between 20 and 50. Varus Flexion Semimembranosus Semitendinosus Biceps femoris Medial gastrocnemius Lateral gastrocnemius Gracilis Sartorius Extension Vastus medialis* Rectus femoris* Rectus femoris* Tensor fascia latae * Note the ability of the quadricep muscles to produce both valgus and varus moments, owing to their common insertion into the patella tendon. Vastus lateralis* Tensor fascia latae Vastus medialis Vastus lateralis Sartorius Lateral gastrocnemius Gracilis Medial gastrocnemius Valgus Internal Rotation Semimembranous Semitendinosus Biceps femoris External Rotation

120

Official Journal of the American College of Sports Medicine

http://www.acsm-msse.org

FIGURE 1The effective axis of rotation for muscle moment arms shifts between medial and lateral condyles of the knee depending on the externally applied varus or valgus moment. Note the ability of the quadriceps muscles to produce both varus and valgus moments, owing to their common insertion point and the midline attachment of the patella tendon.

2. During PP sidestepping, activation of medial and external rotation muscles will increase compared to the Run task, to counter the applied valgus and internal rotation moments at the knee. 3. During the PP crossover cut, activation of lateral and internal rotation muscles will increase compared to the Run task, to counter the applied varus and external rotation moments at the knee. 4. During UN conditions, the CNS will adopt a strategy of generalized co-contraction, as opposed to directed activation patterns to counter the VV and IE loads at the knee. 5. Average muscle activation during the UN cutting tasks will be greater than that found in the PP cutting condition to counter the large increase in VV and IE loads.

METHODS
The experiments performed have been presented previously (3,4) but will be described briefly for the sake of clarity. Subjects. Eleven healthy male soccer players, without history of lower limb injury, volunteered for this investigation (mean SD age: 21.3 3.4 yr; height: 179.4 7.8 cm; mass: 74.1 7.1 kg). Before data collection, the testing protocols were explained to the subjects and informed, written consent obtained to comply with the ethics requirements of the University of Western Australia. Experimental design. Subjects were asked to perform repeated trials of four tasks in the gait laboratory during three separate sessions. The first session was conducted to familiarize the subjects with the four different tasks, and, hence, data were only collected in sessions 2 and 3. The four tasks included a straight run (RUN), sidesteps to 30 (S30) and 60 (S60) from the direction of travel (stepping left off the right foot), and a crossover cut to 30 (XOV) from the direction of travel (cutting right off the right foot). Infrared timing gates were used to monitor the approach running speed, which was delimited to 3.0 ms1 (~10 kmh1). All tasks were performed in random order to account for fatigue, with 10 trials of each maneuver recorded (5 PP and 5 UN, see below), giving a total of 40 trials per session. One-minute intervals were also given between each trial to reduce the effects of fatigue. Subjects performed the tasks in bare feet to negate any confounding effects that different footwear might have on the muscle activity surrounding the knee joint. Muscle activation patterns to perform a specific task may also be affected by learning, particularly if the task in question is novel. To reduce any effects that learning may have on the muscle activation patterns, we chose tasks that would be familiar to a group of soccer players and ensured that each subject performed the familiarization trial before testing sessions 2 and 3. The trials were randomly assigned to PP and UN conditions by using a target board mounted with a set of light emitting diodes (LED). For all PP conditions, the appropriate LED on the target board was turned on at the beginning of the approach run. For the UN conditions, an LED was turned on so that the subject was required to make the
Medicine & Science in Sports & Exercise

(1,15,16). In the case of other muscles surrounding the knee joint, if one assumes rotation about one condyle due to an external load, then the resulting motion will increase the moment arms of antagonist muscles and reduce the moment arms of agonist muscles. For example, an external valgus load may cause rotation of the tibia about the lateral condyle of the femur. In doing so, the moment arm of the medial hamstring muscles to produce a varus moment increases while the moment arm of the lateral hamstring muscles to produce a valgus moment reduces. This mechanism simplifies the task of the CNS to stabilize the joint in the frontal plane, as muscle forces do not have to be anticipated or adjusted according to the external moment applied to the joint. Rather, the CNS can employ a more general strategy of co-contraction and joint stability can be maintained by changes in the muscle moment arms alone. The purpose of this article was to investigate the activation patterns of muscles surrounding the knee during PP and UN running and cutting tasks, with respect to the external moments applied to the joint. Three fundamental questions were posed. First, does the CNS modify activation patterns to stabilize the knee for the increases seen in VV and IE loading during the cutting tasks, compared with the run, even when the flexion moments were essentially the same? Second, does the CNS use a directed activation of medial/ lateral and internal/external rotation muscle groups according to the expected moments experienced at the knee during PP cutting tasks, or does the CNS adopt a general cocontraction strategy? Third, does the CNS adopt a more generalized co-contraction pattern of muscle activation, compared with a directed muscle activation strategy, during cutting tasks that are unanticipated compared to when these tasks are preplanned? Given these fundamental questions and prior knowledge of the external loads placed on the knee joint, we hypothesized the following: 1. Average muscle activation will increase in the cutting tasks compared with the straight run (RUN), in line with the increased VV and IE loads.
MUSCLE STRATEGIES DURING RUNNING AND CUTTING

121

decision on which task to perform just before reaching a force plate. Subjects were required to perform a trial session 1 wk before data collection to obtain appropriate delay times for each individual and reduce potential effects of learning. Data collection. Electromyographic (EMG) data were recorded at 2000 Hz for 3 s from a 10-channel EMG system (Motion Lab Systems, Baton Rouge, LA) and collected synchronously with motion and force plate data with a VICON motion analysis system (Oxford Metrics, Oxford, UK). Bipolar surface electrodes (Ag-AgCl 3 M Red Dot) were placed on the following 10 muscles: semimembranosus (semimem), biceps femoris (bifem), sartorius (sar), tensor fascia latae (tfl), gracilis (gra), vastus lateralis (vaslat), vastus medialis (vasmed), rectus femoris (rf), medial gastrocnemius (medgas), and lateral gastrocnemius (latgas). Surface electrode placements and preparation were in accordance to Delagi et al. (11). The skin was shaved and exfoliated using course plastic gauze then cleaned with alcohol before electrode placement. Amplifier gains for each muscle were adjusted such that the recorded signal was maximal without clipping of data. Three-dimensional lower limb kinematics and kinetics were determined using a six-camera 50-Hz VICON motion analysis system in conjunction with force plate data collected at 2000 Hz. For the purpose of this study, summarized measures of knee joint moments is given to facilitate the interpretation of the activation data (Table 1). Full examinations of joint moment data during PP and UN running and cutting have been published previously (3,4). Data analysis. Raw EMG data were high pass filtered using a zero-lag 4th-order recursive Butterworth filter (cutoff frequency of 30 Hz) to remove movement artifact, then full-wave rectified and low-pass filtered using a zero-lag 4th-order Butterworth filter with a cut-off frequency of 6 Hz. The muscle activation profiles were then normalized to the RUN task in the following way: 1. The maximum peak filtered EMG for each muscle obtained over five preplanned RUN trials were obtained. 2. The average peak filtered EMG for each muscle was calculated over the five trials. 3. EMG data from all trials were then divided by the average peak filtered EMG for each muscle. Branch et al. (5) showed that normalizing EMG to a functional task reduced intersubject variability compared with normalizing to a maximum voluntary contraction. Neptune and colleagues (18) and Rand and Ohtsuki (20) have used a similar approach to normalize EMG signals during dynamic cutting tasks. In our previous articles, we analyzed the knee joint moments in the weight acceptance (WA), peak push-off (PPO), and final push-off (FPO) phases. In this article, the activation in FPO was not analyzed because 1) the very small loading magnitudes experienced in this phase would pose very little risk of ligament injury (3,4), and 2) the electromechanical delay between EMG and muscle force production means that the activation in FPO, being approximately the last 35 ms1 of stance, would not correspond to the loading experienced in this phase. The greatest loads, and 122
Official Journal of the American College of Sports Medicine

FIGURE 2Schematic of the three stages of stance phase, determined using resultant ground reaction force. PC, precontact; WA, weight acceptance; PPO, peak push-off.

hence risk of ligament injury, occurred during WA and PPO; therefore, muscle activation data were analyzed in these phases to compare with the external joint moments. Owing to the preprogrammed nature of muscle activation, we also compared the activation patterns of different tasks and conditions 50 ms1 before foot contact (precontact, or PC). In summary, muscle activation data were averaged over the following stages: 1. Precontact (PC)from 50 ms1 before foot strike to foot strike. 2. Weight acceptance (WA)from foot strike to the first trough in the GRF. 3. Peak push-off (PPO)10% either side of the peak GRF. Average activation of all muscles, as well as the average activation of the flexor and extensor muscle groups, were determined for each phase (see Fig. 2). Muscles were also grouped by their potential to produce varus or valgus and internal or external rotation moments to calculate the average activation of the lateral, medial, internal, and external rotation muscles, respectively (Table 2). Medial muscles were the gra, vasmed, semimem, medgas, and sar, whereas lateral muscles included tfl, vaslat, bifem, and latgas (6). The vasti muscles are capable of resisting both varus and valgus loads at the knee owing to their insertion point into the patella tendon. However, both the vasmed and vaslat have other insertions into the knee capsule, and possibly into the collateral ligaments (9). Hence, it may be possible that the true resultant moment arms for the vasmed and vaslat were more medial and lateral, respectively. Consequently, we classified vasmed as being medial and vaslat as being lateral, as suggested by Buchanan et al. (6). Muscle rf was not included in the analysis of medial/lateral activations due to its midline insertion into the patella. To determine the role of muscles in supporting IE rotation, we compared the activation of semimem and bifem as a ratio (bifem/semimem). We chose to analyze only the hamstring muscles as IE rotators at the knee, owing to the potential of these muscles to produce IE moments, a product of their respechttp://www.acsm-msse.org

tive moment arms and cross-sectional areas (8,12). Muscle semimem has a moment arm to produce internal rotation that is 25 times greater than the vasti and gastrocnemius muscles (8), and a greater cross-sectional area than sar or gra (which have similar internal rotation moment arms). Muscles bifem and tfl are the only external rotators at the knee (8), with bifem having a larger moment arm and cross-sectional area than tfl. A co-contraction ratio (CCR) was determined to define the relative activation of the flexor and extensor muscles crossing the knee. The CCR was calculated the same way that was described in Lloyd and Buchanan (16), being the ratio of average activation for flexors and extensors in each phase. The flexors activation was chosen as the divisor if its value was greater than the extensors and vice versa. Therefore, the CCR was always less than or equal to 1.0. A co-contraction index (CCI) was then calculated for each phase to account for the magnitude of activation. The CCI was the product of the average activation of all muscles and the co-contraction ratio. If we chose a CCR and CCI in which the denominator and numerator were not switched, then the same relative level of co-contraction and average activation of flexors and extensors would, in most cases, produce quite different co-contraction ratios and indices. In addition, if denominator and numerator were not switched when the denominator became small in value (i.e., approaching zero) then the CCR and CCI would approach infinity. The calculation of CCR and CCI used in this article avoided these scenarios. Statistical design. A three-factor repeated measures ANOVA was performed on the average muscle activation data to determine differences in net mean muscle activations across tasks, across phases, and between PP and UN conditions. In addition, a two-factor repeated measures ANOVA was used to determine significant differences between tasks and between PP and UN conditions for the co-contraction data and medial versus lateral muscle groups. Finally, a three-factor repeated measures ANOVA was used to calculated differences in the bifem/semimem ratio across tasks, phases, and PP/UN conditions. All statistical analyses were performed using DataDesk (Data Description Inc., Ithaca, NY), with significance set at P 0.05.

FIGURE 3Percentage increase in average muscle activation during preplanned cutting tasks compared with the RUN. PC, precontact; WA, weight acceptance; PPO, peak push-off; * significantly different to RUN, P < 0.01.

strike (in the PC phase), supporting the notion that activation patterns are preprogrammed. Co-contraction strategies were evident to stabilize the knee joint in VV during PP cutting tasks (Fig. 4). The CCR was similar between tasks or phases in the PP conditions, with significant differences only noted between the RUN and S60 at PC and the RUN and S30 at WA (Fig. 4). At the PC phase, the hamstring activation was greater than the quadriceps; therefore, the hamstring activation was used as the CCR divisor at this phase. At WA and PPO, the quadriceps activation was used as the divisor in the CCR calculation. The magnitude of the muscle activation should also be considered along with the CCR to indicate the level of co-contraction at the joint. For example, the CCI in the PP condition were significantly larger (P 0.05) in the cutting tasks compared with the RUN at WA and PPO (Fig. 5), even though the ratios of flexor/extensor activation were similar. This was due to the increased net activation during the cutting tasks compared with the run. At PC, only the S60 task had significantly greater CCI (P 0.05) than the RUN task (Fig. 5). Selected activation of the medial muscle group was also evident during the PP sidestepping tasks. At PC, the activation of the medial muscle group increased by 33% (S60) and 22% (S30) compared with the lateral muscle group in the PP sidestepping conditions (P 0.05). There was no difference between activation of medial and lateral muscle

RESULTS
Preplanned conditions. Under PP conditions, the average activation of all muscles during the cutting tasks was significantly greater compared with the RUN across all phases (Fig. 3, P 0.01). At each phase of the cutting tasks, there was greater muscle activation than the RUN task. These increases coincide with the increase in the VV and IE knee moments observed in these cutting tasks (4). Compared with the RUN task, the VV and IE loads at the knee joint increased substantially during the cutting tasks, with combined external loads of valgus/internal rotation during sidestepping and varus/external rotation during the crossover cut (4). Muscle activation also increased before heel
MUSCLE STRATEGIES DURING RUNNING AND CUTTING

FIGURE 4 Co-contraction ratios (CCR) of flexor and extensor muscles during all preplanned tasks. PC, precontact; WA, weight acceptance; PPO, peak push-off; * significantly different to RUN, P < 0.05. Medicine & Science in Sports & Exercise

123

FIGURE 5Co-contraction indices (CCI) during all preplanned tasks. PC, precontact; WA, weight acceptance; PPO, peak push-off; * significantly different to RUN, P < 0.05.

FIGURE 7Percentage change in muscle activation during unanticipated conditions compared with preplanned conditions. * significantly different to preplanned condition, P < 0.05.

groups during the RUN and XOV tasks, suggesting the use of a generalized co-contraction pattern in these tasks. Selected activation of bifem and semimem was also found during the PP cutting tasks, as evident from a change in the bifem/semimem ratio between tasks. At PC, the bifem/semimem ratio was ~0.7 for the S60, S30, and RUN tasks, indicating that semimem was more active than bifem before foot contact (Fig. 6). The bifem/semimem ratio for the XOV task was significantly greater than the RUN task at PC (P 0.05), with equal activation of bifem and semimem (ratio of ~1.0). The bifem/semimem ratios increased across all trials during WA (P 0.01), such that bifem was activated 1.31.9 times more than semimem. The highest relative activation of bifem was measured during the S60 task, followed by the S30, the RUN, and then the XOV (Fig. 6). However, only the S60 bifem/semimem ratio was significantly different to the RUN task (P 0.05), although there was also a trend for S30 task to be greater. During PPO, bifem/semimem ratio was greater than in WA (P 0.05). Similarly to WA, at PPO the S60 task was the only task to display a significant difference to the RUN (P 0.05), although there was also a trend for S30 task to be greater (Fig. 6). Unanticipated conditions. When performed in the UN condition, the net average muscle activation increased between 10 and 25% for the sidestepping tasks (P 0.05), with the greatest increase occurring in PC (Fig. 7). The UN condition had little effect on the muscle activation used to perform the RUN task and the only effect the UN condition

had on the XOV task was a 6% decrease in activation at WA (P 0.05). There were no significant changes in the CCR when the cutting tasks were performed in the UN condition, indicating that the ratio of flexor and extensor muscle activation remained the same between conditions. However, the slight increase in average muscle activation during the UN S60 task meant that there was a proportional increase in the CCI at PC and WA of 19% and 15%, respectively (P 0.05). There was also an increase in CCI for the UN S30 task at PC of 28% (P 0.001). The CCI was similar between conditions for the RUN and XOV tasks. There were no significant differences between medial and lateral muscle activations during any of the UN cutting tasks, in any phase of the maneuvers. Differences were also less noticeable when comparing the bifem/semimem ratios between the tasks in the UN condition. At PC, the bifem/ semimem ratios were similar for all tasks (Fig. 8) and comparable to values observed in the PP condition at PC (Fig. 6). At WA, the bifem/semimem ratios were not as large as those seen in the PP condition, reduced by 21%, 26%, and 31% for the S60, S30, and XOV tasks, respectively (Fig. 8). This may be viewed as a tendency to have more equal activation of bifem and semimem, i.e., the bifem/semimem ratio was closer to 1.0. The activation of bifem was much greater than semimem at PPO compared with PC or WA, and was similar to the bifem/semimem ratio measured in PPO during the PP condition (Fig. 8). The bifem/semimem ratio was similar between the RUN and sidestepping tasks (Fig. 8). The XOV task had the lowest

FIGURE 6 bifem/semimem ratios during preplanned conditions for all tasks across each phase; * significantly different to RUN task at similar phase, P < 0.05; bifem/semimem ratios at WA were significantly greater than those at PC, P < 0.01; bifem/semimem ratios at PPO were significantly greater than those at WA, P < 0.05.

FIGURE 8 bifem/semimem ratios during unanticipated conditions for all tasks across each phase. * significantly different to RUN task at similar phase, P < 0.05; bifem/semimem ratios at PPO were significantly greater than those at PC or WA, P < 0.05. http://www.acsm-msse.org

124

Official Journal of the American College of Sports Medicine

bifem/semimem ratio at PPO and was ~40% less than the RUN and sidestepping tasks (P 0.05) (Fig. 8).

DISCUSSION
The purpose of this article was to investigate the activation patterns of muscles surrounding the knee during PP and UN running and cutting tasks with respect to five specific hypotheses. The following discussion answers these hypotheses and examines the general questions posed in the introduction. In answer to the first hypothesis, it appears that the CNS activates the knee muscles to stabilize the joint in VV and IE other than just FE during dynamic tasks. Besier et al. (4) showed that, compared with the RUN, VV and IE moments increased in cutting tasks, whereas FE moments did not. The activation of all muscles increased at PC, WA, and PPO during the cutting tasks (Fig. 3) compared with the RUN, which may have been directed to stabilize the knee as it experienced larger VV and IE moments. This response has been shown for isometric knee loading where activation patterns used to stabilized the knee in VV superimposed on the activation patterns to generate FE moments (1,7,16). Cowling and Steele (10) suggested hamstring activation patterns before initial contact in land and stop tasks are preplanned to counter ACL loading at landing. The present study supports this general finding that activation patterns are preplanned to counter loading in VV and IE directions. At PC the level of activation was scaled in anticipation of the size of expected VV and IE moments and not the FE moments. Our results show that the magnitude of the average muscle activation during the PP tasks was larger when greater VV and IE moments were expected (Fig. 3). For example, the S60 elicited greater valgus and internal rotation moments as well as greater muscle activation than the S30 task (see Figs. 4 and 5 in Besier et al. 4). Similarly, the combined varus/external rotation loads during the XOV were much greater than the varus/external loads during the RUN, with a corresponding increase in muscle activation. Toussaint et al. (21) found anticipatory adjustments to applied joint moments were scaled by the expected magnitude of the perturbation, and Benvenuti et al. (2) also found the scaling of postural muscle activation varied with the magnitude of the destabilizing forces imposed by an expected perturbation. This current work extends these findings to suggest that the CNS scales the activation level for stabilization of the knee joint in directions other than the prime motion of flexion and extension during dynamic tasks. In answer to hypotheses 2, 3, and 4, it appears that combinations of co-contraction and selected muscle activation are present during dynamic cutting tasks. During the PP sidestepping tasks, there was selective activation of medial muscle groups to counter the external valgus load applied to the joint (4) and selected activation of bifem to counter the applied internal rotation moments (4), supporting the second hypothesis. However, co-contraction of flexors and extensors also increased during PP sidestepping tasks. Lloyd and Buchanan (16) also observed combinations of selected muscle activation strategies and co-contraction of flexor and
MUSCLE STRATEGIES DURING RUNNING AND CUTTING

extensor muscles to support isometric VV loads at the knee. Using both of these strategies, the CNS is capable of altering a set of activation patterns at the knee to suit the external moments applied to the joint during cutting tasks. These findings support previous research showing increased lower limb co-contraction with tasks that require high levels of stability (e.g., Llewellyn et al. 13). During the XOV task where large varus moments were experienced, posing a possible risk of lateral collateral ligament injury (3), a generalized co-contraction strategy was used to stabilize the knee joint, with few differences found between medial/lateral muscle activation patterns and bifem/semimem ratios, thus not supporting our third hypothesis. Toussaint et al. (21) showed that previous experience in performing a task altered preprogrammed postural movements required to perform that skill. Perhaps the XOV task was less familiar to subjects than the sidestepping tasks; therefore, the muscle activation patterns required to counter the varus and external rotation loads during this maneuver may not be as finely tuned as those chosen to perform the sidestepping tasks. This raises the question, can be people be trained to improve muscle activation patterns to counter the external loads applied to the knee joint during cutting maneuvers to reduce the risk of ligament injury? Perhaps with further training or practice performing the XOV task, muscle activation patterns might be redirected to counter the applied varus and external rotation load. With regard to the fourth hypothesis, a generalized cocontraction strategy was adopted in preference to selected activation of medial/lateral muscle groups or bifem/semimem ratios in the performance of the UN sidestepping maneuvers. The ratio of flexor/extensor muscle activation (CCR) was similar between anticipatory conditions, but the increase in average muscle activation during the UN condition accounted for an increased level of co-contraction (noted by the increased CCI). The final hypothesis was in part supported by the observed increase in muscle activation during the UN sidestepping tasks compared with the PP condition. However, the relative increase in muscle activation was small (10 20%) when compared with an increase in valgus moments of ~70% and an increase in internal rotation moments of ~90% during the UN sidestepping tasks at WA (Table 1). The varus moments during the UN XOV task increased by ~60% and the external rotation moments increased by ~100% compared with the PP condition (Table 1), but there was actually a decrease in the net activation during WA and PPO, which did not support the final hypothesis. This mismatch between the changes in the preprogrammed muscle activation strategy and the increased VV and IE moments applied to the joint during unanticipated tasks may be responsible for placing large loads on knee joint ligaments. Although there are other factors involved in the etiology of knee ligament injury, such as joint position/geometry and ligament strength, it appears as though the risk of injury during a noncontact cutting maneuver increases under UN conditions. However, ligament loading needs to be examined under these dynamic condiMedicine & Science in Sports & Exercise

125

tions, taking into account the force produced by muscles, to adequately assess the potential for knee ligament injury. It is important to note that the different activation patterns at the knee may also have been due to moment requirements at the ankle and hip joints. The biarticular muscles of hip and knee, and ankle and knee, contribute to the moments generated at the hip and ankle, respectively. Thus, if changes occurred in the external moments at the hip and ankle in the different maneuvers, or between the PP and UN conditions, these may have resulted in changed activation patterns of the biarticular muscles. Consequently, the activation of uniarticular muscles of the knee would have to compensate for the altered activation of the biarticular muscles to maintain a moment balance at the knee. However, Lloyd and Buchanan (15) showed that in static VV loading of the knee, the activation patterns of biarticular muscles at the knee appeared to be selected for stabilization of the knee independent of their role at the hip. Two limitations to this study should be considered before generalizing the main findings to noncontact knee ligament injuries in sport. First, the subjects performed the tasks in bare feet, which is uncommon in most game situations. As stated previously, the bare-feet condition was used to account for any confounding effects that may result from different shoe/surface interactions. Performing the tasks in bare feet may be similar to changing direction in wet and slippery conditions, which may be an underestimate of the actual joint loads experienced by athletes wearing cleats on dry ground. A much more rigorous study is required to investigate the interaction between the traction provided by a specific shoe/surface combination and the corresponding joint loads, kinematics, and muscle activation patterns at the knee. The second limitation was the designated running speed of 3 ms1, which may or may not be representative of the speed at which ligament injury occurs in sport. Unfortunately, no data were available to represent the actual running speeds at which ligament injury occurs in sport, although a wide range of speeds would be expected, depending on the nature of the sport (netball and basketball compared with soccer and football, for example). As experimental data are unavailable at this time, we can only speculate that the magnitude of the VV and IE loads may increase in faster compared to slower cutting. However, this may not be the case as Novacheck (19) showed that the external flexion moment applied to the knee during sprinting at 4 ms1 was actually ~10% less than that found in running at 3 ms1. More importantly, however, athletes perform running and cutting maneuvers from slow to fast speeds without ligament injury. The question should then be asked, what is so different about the one sidestep that causes an injury? The comparison of the UN to the PP REFERENCES
1. ANDRIACCHI, T. P., G. B. ANDERSSON, R. ORTENGREN, and R. P. MIKOSZ. A study of factors influencing muscle activity about the knee joint. J. Orthop. Res. 1:266 275, 1984. 2. BENVENUTI, F., S. J. STANHOPE, S. L. THOMAS, V. P. PANZER, and M. HALLETT. Flexibility of anticipatory postural adjustments revealed

performance of the maneuvers, as examined in this article and our previous article (4), is one possible mechanism by which alterations to technique may increase the risk of injury. It is obvious that further investigations of the link between technique and risk of injury are required. If the ultimate goal of this research is to reduce the likelihood of noncontact knee ligament injuries in sport, without compromising the performance of the task, there appears to be a few options for the preparation of athletes. First, reducing the external load applied to the knee joint will reduce the potential for ligament loading. Reducing the external load could possibly be achieved by altering the kinematics to perform the maneuver (3,14); however, the techniques to accomplish this need further study. Second, improving ones ability to anticipate forthcoming movements will serve to reduce the external load placed on the joint (3,14) and allows the CNS time to alter the preprogrammed activation patterns used to perform the task. This study has shown that under PP conditions, the CNS is capable of scaling muscle activation patterns in line with the increasing VV and IE loads in the cutting tasks. However, under UN conditions, there appears to be a mismatch between the activation patterns and the external load applied to the joint where only modest increases in net activation occurred with large increases in VV and IE loads. Finally, appropriate training or practice may alter preprogrammed muscle activation patterns used to perform a task, and should focus on improving selected activation and/or increasing co-contraction of flexors and extensors.

SUMMARY
The CNS alters muscle activation patterns at the knee to perform PP sidestepping and crossover cutting tasks, in anticipation of the external loads placed on the joint. Selected activation of medial muscles and bifem was found during the sidestepping tasks when valgus and internal rotation moments were applied to the knee, respectively. Co-contraction of flexor and extensor muscle groups was also evident during the PP cutting tasks. When the cutting maneuvers were performed in UN conditions, a generalized co-contraction strategy was predominantly used. Muscle activation levels only increased by 10 20% when the maneuvers were unanticipated, and it is doubtful whether this increase is enough to offset the 70% and more increases in VV and IE joint loads. As such, there appears to be a mismatch between the activation patterns and external load applied to the joint under UN conditions, which may play an important role in the etiology of noncontact knee ligament injuries.
We would like to acknowledge the Australian Football Research and Development Board for their financial support.

by self-paced and reaction-time arm movements. Brain Res. 761: 59 70, 1997. 3. BESIER, T. F., D. G. LLOYD, T. R. ACKLAND, and J. L. COCHRANE. Anticipatory effects on knee joint loading during running and cutting maneuvers. Med. Sci. Sports Exerc. 33:1176 1181, 2001.
http://www.acsm-msse.org

126

Official Journal of the American College of Sports Medicine

4. BESIER, T. F., D. G. LLOYD, J. L. COCHRANE, and T. R. ACKLAND. External loading of the knee joint during running and cutting maneuvers. Med. Sci. Sports Exerc. 33:1168 1175, 2001. 5. BRANCH, T. P., R. HUNTER, and M. DONATH. Dynamic EMG analysis of anterior cruciate deficient legs with and without bracing during cutting. Am. J. Sports Med. 17:35 41, 1989. 6. BUCHANAN, T. S., A. W. KIM, and D. G. LLOYD. Selective muscle activation following rapid varus/valgus perturbations at the knee. Med. Sci. Sports Exerc. 28:870 876, 1996. 7. BUCHANAN, T. S., and D. G. LLOYD. Muscle activation at the human knee during isometric flexion-extension and varus-valgus loads. J. Orthop. Res. 15:1117, 1997. 8. BUFORD, W. L., F. M. IVEY, T. NAKAMURA, R. M. PATTERSON, and D. K. NGUYEN. Internal/external rotation moment arms of muscles at the knee: moment arms for the normal knee and the ACLdeficient knee. Knee 8:293303, 2001. 9. CALLIET, R. Knee Pain and Disability. Philadelphia: F. A. Davis, 1992, pp. 20 26. 10. COWLING, E. J., and J. R. STEELE. Is lower limb muscle synchrony during landing affected by gender? Implications for variations in ACL injury rates. J. Electromyogr. Kinesiol. 11: 263268, 2001. 11. DELAGI, E. F., A. PEROTTO, J. IAZZETTI, and D. MORRISON. Anatomic Guide for the Electromyographer. Springfield, IL: Charles C Thomas, 1982, pp. 141207. 12. KAPANDJI, I. A. The Physiology of the Joints. Edinburgh: Churchill Livingstone, 1970, p. 132.

13. LLEWELLYN, M., J. F. YANG, and A. PROCHAZKA, A. Human Hreflexes are smaller in difficult beam walking than in normal treadmill walking. Exp. Brain Res. 83:2228, 1990. 14. LLOYD, D. G. Rationale for training programmes to reduce ACL injuries in Australian Football. J. Orthop. Sports Phys. Ther. 31:645 654, 2001. 15. LLOYD, D. G., and T. S. BUCHANAN. A model of load sharing between muscles and soft tissues at the human knee during static tasks. J. Biomech. Eng. 118:367376, 1996. 16. LLOYD, D. G., and T. S. BUCHANAN. Strategies of muscular support of varus and valgus isometric loads at the human knee. J. Biomech. 34:12571267, 2001. 17. MARKOLF, K. L., D. M. BURCHFIELD, M. M. SHAPIRO, M. F. SHEPARD, G. A. FINERMAN, and J. L. SLAUTERBECK. Combined knee loading states that generate high anterior cruciate ligament forces. J. Orthop. Res. 13:930 935, 1995. 18. NEPTUNE, R. R., I. C. WRIGHT, and A. J. VAN DEN BOGERT. Muscle coordination and function during cutting movements. Med. Sci. Sports Exerc. 31:294 302, 1999. 19. NOVACHECK, T. F. The biomechanics of running. Gait Posture 7:7795, 1998. 20. RAND, M. K., and T. OHTSUKI. EMG analysis of lower limb muscles in humans during quick change in running directions. Gait Posture 12:169 183, 2000. 21. TOUSSAINT, H. M., Y. M. MICHIES, M. N. FABER, D. A. COMMISSARIS, and J. H. VAN DIEEN. Scaling anticipatory postural adjustments dependent on confidence of load estimation in a bi-manual whole-body lifting task. Exp. Brain Res. 120:8594, 1998.

MUSCLE STRATEGIES DURING RUNNING AND CUTTING

Medicine & Science in Sports & Exercise

127