Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
MECHANISM AND PREVENTION OF ANTERIOR CRUCIATE LIGAMENT INJURIES IN
SPORT
Cyril J. Donnelly, M.Sc.
This thesis is presented for the degree of Doctor of Philosophy at The University of Western
Australia
The School of Sport Science, Exercise and Health Biomechanics
June, 2012
ii
To my mother and father
iii
Abstract
Review of the anterior cruciate ligament (ACL) injury prevention literature has shown
that exercise/training can be used to reduce ACL injury risk and injury rates in general
athletic populations. However, a large gap still exists in the literature, with little to no
research testing the effectiveness of these prophylactic training protocols in community
level training environments. Results from this thesis have shown that when
prophylactic training protocols were implemented in community level training
environments; they were not effective in reducing surrogate biomechanical measures
of ACL injury risk like peak knee joint loading and muscular support. We must begin to
better understand the biomechanical mechanisms by which prophylactic training
protocols act if we can more effectively translate positive laboratory based findings to
community level training environments.
To identify these causal mechanisms, we have developed a novel computational
method capable of identifying causal links between an athlete’s whole-body kinematics
and knee joint kinetics during dynamic simulations of human movement. The
generalised kinematic strategy identified during sidestepping, where one half of non-
contact ACL injuries have been shown to occur was to reposition an athlete’s whole-
body centre of mass medially, towards their desired direction of travel. Through the
development and use of these methods, the ability to identify short, concise and
effective training protocols is possible; increasing the probability of translating ACL
focused research into injury prevention practice in community level training
environments.
iv
REFERENCE DISCLAIMER
This PhD dissertation has in part been submitted or accepted for publication in
internationally recognised journals. For the chapters within this thesis that have been
submitted, or accepted for publication, referencing will be as per the individual journal
guidelines. For chapters that have not been submitted for peer review, the referencing
format will be as per the Journal of Biomechanics.
v
EXECUTIVE SUMMARY
A) Chapter 2
Title: An anterior cruciate ligament injury prevention framework: Incorporating the
recent evidence
A comprehensive review of the ACL injury prevention literature shows that
exercise/training can be used to reduce ACL injury risk and injury rates in general
athletic populations. Though a rationale to use various exercise protocols to reduce
ACL injuries is established, the mechanisms by which it acts are relatively unknown.
Using the six stage injury prevention model to ‘Translate Research into Injury
Prevention Practice’ (TRIPP model), an injury prevention framework specific to, and
detailed for non-contact ACL injuries was developed. Additionally an empirically based
rationale for the design of ACL injury prevention training protocols was also developed.
Within our ACL injury prevention framework, we used a multidisciplinary approach to
develop a model for the aetiology of ACL injuries, and in turn appropriate
countermeasures to reduce injury risk. From previously published empirical research,
three biomechanically based countermeasures were identified:
1) Reduce the magnitude of externally applied flexion, valgus and internal rotation
knee moments during the weight acceptance phase of sidestepping or single-
leg landing.
2) Increase muscular support against these aforementioned joint moments.
3) Increase knee flexion angle and the neuromuscular control of the hip during the
weight acceptance phase of sidestepping and single-leg landing.
Previous literature has shown that the combined effects of plyometric, balance,
resistance and/or technique training are effective in reducing the biomechanical risk
factors associated with ACL injury in ‘ideal’ training environments. However, a large
gap exists in the literature, where little to no research has tested the effectiveness of
these prophylactic training protocols in ‘real-world’ training settings. It is then unknown
if positive laboratory based biomechanical training outcomes can be translated to
community level training environments. Additionally, it is evident that the use of
feedback within this framework is needed to determine how biomechanical factors, like
joint loading and muscle support are targeted following a given training intervention. It
is by identifying these causal links that more effective and targeted ACL injury
vi
prevention training programs can be developed and in turn lead to reduced ACL injury
rates in the future.
The overall goal of this thesis is to begin filling these gaps and determine if positive
laboratory based findings can be transferred to ‘real-world’ community level training
environments. Additionally, we have developed novel computational methods to
identify causal relationships between an athlete’s sidestepping and single-leg landing
techniques, knee joint loading and ACL injury risk. Through this approach, better
injury prevention protocols targeting the biomechanical factors associated with ACL
injury can be developed; transferring positive laboratory based training effects to ‘real-
world’ training environments, and in turn reduce ACL injury rates in community level
training environments.
B) Chapters 3 & 4
Titles:
Part 1 – Changes in knee joint biomechanics following balance and technique training
and a season of Australian Football
Part 2 – Changes in muscle activation following balance and technique training and a
season of Australian Football
Purpose: Determine if balance and technique training (BTT) implemented adjunct to
normal Australian football (AF) training reduces external knee loading and influences
the activation of muscles crossing the knee during sidestepping. Also, determine if an
athlete’s knee joint biomechanics and muscle activation changes over a season of AF.
Finally, determine if changes in muscle activation were proportional to changes in knee
joint loading.
Methodology: 1,001 males volunteered to participate in either 28 weeks of BTT or
‘sham’ training (ST), adjunct to their normal pre-season and regular training. A subset
of 34 athletes (BTT, n = 20; ST, n = 14) were randomly recruited for laboratory-based
biomechanical testing in weeks -1 to 7 and 18 to 25 of the 28 week training
intervention. During biomechanical testing, participants completed a series of running,
pre-planned (PpSS) and unplanned sidestepping (UnSS) tasks. During PpSS and
UnSS, knee joint kinetics in three degrees of freedom and knee flexion kinematics were
calculated from all 34 athletes. Directed co-contraction ratios (DCCR) in three degrees
of freedom and total muscle activation (TMA) during PpSS and UnSS were attained
from 28 (BTT, n = 12; ST, n = 16) of the 34 athletes. A linear mixed model (α = 0.05)
vii
was used to determine if knee joint kinematics, kinetics and muscle activation during
PpSS and UnSS were influenced by 28 weeks of BTT and/or a season of AF.
Results: The main findings from these studies were that BTT, when implemented
adjunct to normal ‘real-world’ AF training, was not effective in reducing a player’s knee
joint kinematics, external knee loading or changing the activation of the muscles
crossing the knee during PpSS and UnSS. However, significant within season training
effects were observed. Peak internal rotation knee moments during PpSS significantly
decreased (p = 0.025) by 45% over a season of AF, while peak valgus knee moments
during UnSS significantly increased (p = 0.022) by 31%. Additionally, significant
increases in knee extensor (p = 0.023) and semimembranosus (p = 0.006) muscle
activation were observed during both PpSS and UnSS. However, TMA was lower
during UnSS when compared with PpSS, even in the presence of significantly elevated
valgus knee moments.
Conclusions: BTT was not effective in changing an athlete’s knee joint biomechanics
or muscle activation during sidestepping when conducted in ‘real-world’ training
environments. Following a season of AF, athletes are better able to support both
frontal and sagittal plane knee loading during PpSS and UnSS. Knee joint
biomechanics respond to normal AF training differently during pre-planned and
unplanned sidestepping. Both pre-planned and unplanned sport tasks are therefore
recommended when assessing the effectiveness of prophylactic training protocols.
Elevated valgus knee loading combined with relatively low TMA during UnSS following
a season of AF suggests an athlete may be at increased risk of ACL injury when
conducting unplanned sports tasks in the latter half of a playing season.
Significance: This is the first series of studies to implement a prophylactic training
protocol in a ‘real-world’ community level training environment with the goal of reducing
the biomechanical factors associated with ACL injury risk. It is clear from these results
that much work is needed before positive laboratory based findings can be translated
to community level training environments. However, the training and biomechanical
testing framework used in this study may help refine future ACL injury prevention
training programs focused on reducing ACL injury risk in community level athletes.
It is also apparent we must begin understanding the biomechanical mechanisms by
which training influences ACL injury risk factors like knee joint kinematics, external joint
loading and muscle support during sidestepping. With causal information available and
the underlying biomechanical factors understood; the development of short, concise,
viii
effective ACL injury prevention countermeasures can be developed. Through this
approach, the probability a community level athlete will adhere to a prophylactic
training protocol is increased, which may lead to reductions in ACL injury rates in the
future.
C) Chapters 5 & 6
Titles:
1. An open-source computational method to optimise simulated human motion to
reduce valgus knee loading during sidestepping and single-leg landing.
2. Optimizing whole-body kinematics to minimise valgus knee loading during
sidestepping: Implications for ACL injury risk
Purpose:
1. Using the Residual Reduction Algorithm (RRA) in the musculoskeletal
modelling software OpenSim, develop a method to optimise a simulation’s
kinematics to minimise peak valgus knee loading during unplanned
sidestepping and single-leg landing.
2. Using these computational methods, identify causal relationships between an
individual’s whole-body kinematics and peak valgus knee moments during the
weight acceptance phase of unplanned sidestepping.
Methodology:
1. A single full-body, 37 degree-of-freedom (DoF) skeletal model in OpenSim was
used to create a dynamic simulation of single-leg landing (SLL) and unplanned
sidestepping (UnSS). The stance limb for each simulation was the right leg.
The RRA in OpenSim and an outer-level optimisation method was used to
create dynamically consistent simulations of sidestepping and SLL during the
weight acceptance phase of stance. Peak valgus knee torque were reduced in
the dynamically consistent simulations of UnSS and SLL, and RRA run again to
produce optimised kinematic solutions with reduced peak valgus knee torque.
ix
2. Nine independent simulations of UnSS were created using the aforementioned
RRA methods, where valgus knee loading was minimised during the weight
acceptance phase of stance.
Results:
1. Using RRA and the outer-level optimisation method, dynamically consistent
simulations of UnSS (peak RMS kinematic errors < 3.0°; residual errors < 2N
and 1Nm) and SLL (peak RMS kinematic errors < 4.0°; residual errors < 1N and
1Nm) were created.
When reducing the maximum allowable valgus joint torque in the dynamically
consistent simulations of UnSS and SLL and RRA ran again, peak valgus knee
torques were reduced by 50% (77.9 Nm) and 26% (23.3 Nm) respectively.
The kinematic changes corresponding to the reduction in valgus knee torque
during UnSS were trunk rotation towards the desired direction of travel (2.9°),
right shoulder adduction (15.7°), left shoulder flexion (4.1°) and right hip
abduction (3.1°) (stance limb, right leg). The kinematic changes corresponding
to the reduction in valgus knee torque during SLL were left hip (7.8º) and knee
(19.3º) extension (stance limb, right leg).
2. Pre-to-post kinematic optimisation, mean peak valgus knee moments were
significantly reduced by 44.2 Nm (p = 0.045) (n = 9). The generalised
kinematic strategy used by all nine simulations to reduce peak valgus knee
moments and subsequent ACL injury risk during UnSS was to redirect the
whole-body centre of mass medially, towards the desired direction of travel.
Conclusions:
1. An outer-level optimisation method with the RAA in OpenSim can be used to
identify causal links between an individual’s whole-body kinematics and valgus
knee loading during both UnSS and SLL sport tasks
2. Re-directing whole-body centre of mass is identified as a generalised kinematic
strategy to reduce valgus knee loading during the weight acceptance phase of
UnSS.
x
Significance: An open-source method has been developed to established causal links
between whole-body kinematics and knee joint kinetics during dynamic simulations of
human movement. Repositioning whole-body centre of mass medially during UnSS is
a generalised kinematic strategy that can be used to reduce an athlete’s ACL injury
risk. The ability to develop more concise ACL injury prevention training programs for
use in community level training environments is indeed possible.
Thesis impact: Results from this thesis showed that prophylactic training protocols,
when implemented in ‘real-world’ training environment were not effective in reducing
surrogate biomechanical measures of ACL injury risk. Following a critical evaluation of
ACL focused research, it is clear we must begin to better understand the
biomechanical mechanisms by which prophylactic training protocols act if we can more
effectively translate positive laboratory findings to ‘real-world’ training environments.
To identify these causal mechanisms, we have developed a novel computation method
within the open-source musculoskeletal modelling framework OpenSim (simtk.org,
Stanford, CA). Using these methods we were capable of identifying a generalised
kinematic strategy to reduce valgus knee loading during UnSS, which is a complex,
multi-body, dynamic movement with an enormous solution space. These methods
possess great potential within the injury prevention field, as the ability to identify causal
links between an athlete’s kinematics and joint loading during a multitude of high risk
sporting tasks is indeed possible. Through the development and use of these methods
the ability to develop short, concise and effective training protocols is indeed possible,
increasing the probability ACL focused research will be translated to injury prevention
practice in the future.
xi
Table of Contents
Abstract iii
Reference Disclaimer iv
Executive Summary v - x
Acknowledgements xxii
Statement of Candidate Contribution xxiii - xxiv
Chapter 1: Introduction 1 - 13
1.1 Background 1 - 2
1.2 Statement of the problem 2
1.3 Aims and hypotheses 3 - 6
1.4 Limitations 6
1.5 Delimitations 6
1.6 Definition of terms 7 - 9
1.7 List of abbreviations 10 - 11
Reference list chapter 1 12 - 13
Chapter 2: An anterior cruciate ligament injury prevention framework: Incorporating the recent evidence
14 - 46
Abstract 15
2.1 Introduction 15 - 16
2.2 ACL injury prevention framework 17
2.3 Injury surveillance 17
2.4 Mechanical aetiology of ACL injury 17 - 20
2.5 Countermeasures 21 - 23
2.6 Countermeasures: technique and knee loading 21 - 22
2.7 Countermeasures: neuromuscular support 22 - 23
2.8 ACL focused training intervention protocols in sport
23 - 37
2.9 Athlete screening 38
2.10 Summary 38 - 40
Acknowledgements 40
Reference list chapter 2 40 - 46
xii
Chapter 3: Changes in knee joint biomechanics following balance and technique training and a season of Australian football
47 - 64
Abstract 48
3.1 Introduction 48 - 50
3.2 Methods 50 - 56
3.2.1 Participant population – training intervention 50
3.2.2 Participant population – biomechanical testing
50 - 51
3.2.3 Training protocol 51 - 52
3.2.4 Biomechanical testing protocol 53 - 55
3.2.5 Analysis 55 - 56
3.2.6 Statistics 56
3.3 Results 56 - 58
3.4 Discussion 58 - 60
3.5 Conclusions 61
Acknowledgements 61
Competing interest statement 61
Contributor statement 61
Funding statement 62
Reference list chapter 3 62 - 64
Chapter 4: Changes in muscle activation following balance and technique training and a season of Australian football
65 - 80
Abstract 66
4.1 Introduction 66 - 68
4.2 Methods 68 - 72
4.2.1 Participant population – training intervention 68
4.2.2 Participant population – biomechanical testing
68 - 69
4.2.3 Training protocol 69
4.2.4 Biomechanical testing protocol 70 - 71
4.2.5 Analysis 71 - 72
4.2.6 Statistics 72 -73
4.3 Results 73 - 75
xiii
4.4 Discussion 75 - 77
4.5 Conclusions 77
Acknowledgements 78
Competing interest statement 78
Contributor statement 78
Funding statement 78
Reference list chapter 4 79 - 80
Chapter 5: An open-source computational method to optimise simulated human motion to reduce valgus knee loading during sidestepping and single-leg landing
81 - 97
Abstract 82
5.1 Introduction 82 - 84
5.2 Methods 84 - 91
5.2.1 Experimental data collection 84 - 86
5.2.2 Dynamically consistent simulation
86 - 90
5.2.3 Minimisation of valgus knee loading 90
5.3 Results 91 - 93
5.4 Discussion 93-94
5.5 Conclusions 95
Reference list chapter 5 95 - 97
Chapter 6: An open-source computational method to optimise simulated human motion to reduce valgus knee loading during sidestepping and single-leg landing
98 - 117
Abstract 99
6.1 Introduction 99 - 101
6.2 Methods 101 - 106
6.3 Results 107 - 110
6.4 Discussion 111 - 114
Acknowledgements 114
Reference list chapter 6 114 - 117
Chapter 7: Summary and conclusions 118 - 128
7.1 Thesis goals 118 - 119
7.2 Specific aims and hypotheses 119 - 125
xiv
7.2.1 Chapter 2: An anterior cruciate ligament injury prevention framework: Incorporating the recent evidence.
119
7.2.2 Chapter 3: Changes in knee joint biomechanics following balance and technique training and a season of Australian Football.
120
7.2.3 Chapter 4: Changes in muscle activation following balance and technique training and a season of Australian Football
121 - 123
7.2.4 Chapter 5: An open-source computational method to optimise simulated human motion to reduce valgus knee loading during sidestepping and single-leg landing
123 - 125
7.2.5 Chapter 6: An open-source computational method to optimise simulated human motion to reduce valgus knee loading during sidestepping and single-leg landing
125
7.3 Summary of study limitations 126
7.4 Future research 127-128
Reference list chapter 7 129
Appendix A – Training protocols 130
Appendix B – UWA upper and lower body models 131
Appendix C – sEMG analysis software 132
Appendix D – 37 DoF OpenSim full body model 133
Appendix E – OpenSim kinematic export software 134
Appendix F – OpenSim GRF export software 135
Appendix G – Outer-level optimisation software 136
Appendix H – Compare forces pre-to-post optimisation 137
Appendix I – Compare kinematics pre-to-post optimisation 138
Appendix J – avi images pre-to-post optimisation 139
xv
List of Figures
Figure 2.1 Anterior cruciate ligament (ACL) strain of the left leg and vertical ground reaction force (GRF) recorded during the stance phase of a stop-landing sport task. Data was then ensemble average of three stop-landing sport tasks. Foot strike occurs at approximately 23% of cycle. The right foot is placed on the ground at approximately 98% and marks the end of the hop cycle. [Adapted from Cerulli et al. (2003)].
19
Figure 2.2 Relationship between relative elongation of the anteromedial bundle (AMB) left A and posterolateral bundle (PLB) right B relative to knee flexion angle during stance phase of gait. [Adapted from Wu, Hosseini, et al. (2010)].
20
Figure 2.3 Linear trend line for landing ACL strain versus quadriceps pre-activation forces for pool of all knees (peak strains) measured during upward impulse. (Mean ± standard error of the mean). [Adapted from Hashemi et al. (2010)].
23
Figure 2.4 ACL injury prevention framework to translate ACL focused research into injury prevention practice.
39
Figure 3.1 Experimental data flow of training intervention and biomechanical testing sessions 1 and 2. BTT and ST numbers were only reported in testing session two as the biomechanists conducting the data collections were blinded to the training intervention codes of each participant until the statistics phase of the analysis. Mean ± standard deviation age, body mass and height were reported for participants who completed both testing session 1 and 2.
51
Figure 3.2 Above: frontal (1) and transverse (2) view of the sidestep sport maneuvers conducted during biomechanical testing. The solid black lines were used as direction cues for participants during change of direction tasks. Below: mid pelvis position (x, y) coordinates 50 frames prior to heel contact (A), at heel contact (B), contralateral leg heel contact (C) and ipsilateral leg mid swing (D) were used to define vectors AB and CD. The cosine of the dot product between vectors AB and CD represents a participants CoD angle during sidestepping.
54
Figure 4.1 Experimental data flow of training intervention and biomechanical testing sessions 1 and 2. BTT and ST numbers were only reported in testing session two as the biomechanists conducting the data collections were blinded to the training intervention codes of each participant until the statistics phase of the analysis. Mean ± standard deviation age, body mass
69
xvi
and height were reported for participants who completed both testing session 1 and 2.
Figure 5.1 Frontal view of the SLL procedure. Frame 1: participant jumps with preferred jumping leg. Frame 1-4: the ball is swung laterally away from their preferred jumping leg, while the participant is in the flight phase. Frame 8 participant lands with preferred jumping leg on a 1.2x1.2m force platform.
86
Figure 5.2 The subject in this study was a male WAAF player. (a) Movement analysis data, including full body, three-dimensional marker trajectories and GRF, were collected during overground straight-line running. (b) A dynamic simulation of the subject was created using a three-step process: 1) a musculoskeletal model with 37 degrees of freedom driven by 37 actuators was scaled to the participant’s joint centres and total body mass; 2) inverse kinematics determined values of the model’s generalised coordinates from the experimentally recorded kinematic data; and 3) RRA was used to produce an optimal set of excitations that produced a dynamically consistent simulation (Equation 2). Note: an outer-level optimisation (Equation 3) determined input parameters for the inner-level optimisation (RRA) to generate the dynamically consistent simulation.
89
Figure 5.3 Largest differences ordered by decreasing magnitude for (a) kinematic errors (accelerations integrated twice), (b) residual forces/torques, and (c) joint torques resulting from simulations generated using RRA as defined by a typical users intuition (blue, before) and then by the outer-level optimisation method (red, after). Also displayed are 10 of the 74 input parameters chosen by a typical user’s intuition (blue, before) and the outer-level optimisation method (red, after). These input parameters include kinematic tracking weights (d), maximum residual forces/torques (e), and (f) maximum joint torques.
91
Figure 5.4 Peak flexion, valgus and internal rotation knee moments pre-to-post kinematic optimisation calculated during the WA phase of sidestepping (Left) and SLL (Right).
92
Figure 6.1 Overview of the experimental procedure: motion data collection (A), skeletal modelling and residual reduction (B) and optimisation WB kinematics to minimised peak valgus knee moments (C).
101
Figure 6.2 Depiction of 37 DoF, 14 segment full-body rigid-linked skeletal model. The pelvis segment with respect to ground was defined using 3 translations and 3 rotations (6 DoF). A ball-and-socket was used to represent the hip, shoulder and pelvis to trunk/head joints (3 DoF). The wrists were modelled
103
xvii
as universal joints (2 DoF). The radial-ulnar, elbow and ankle joints were modelled as revolutes (1 DoF). The knee joint (3 DoF) was modelled as a planar joint in the flexion/extension axis which allowed the tibia to translate as a function of knee flexion angle (Delp et al., 1990); internal/external rotation and abd/adduction were modelled as universal joints.
Figure 6.3 Kinematic mapping of a typical simulation representing the absolute kinematic changes (q) from pre-to-post kinematic optimisation for all DoF within the skeletal model (N = 37) at 20% intervals during WA of UnSS.
106
Figure 6.4 Peak mean knee flexion, valgus and internal rotation moments pre-to-post kinematic optimisation calculated during the WA phase of an UnSS. Symbol * indicates a significant change over time (α = 0.05).
107
Figure 6.5 Mean peak changes in WB CoM relative to stance foot CoM position pre-to-post kinematic optimisation. Anterior and medial changes are towards the desired change of direction pathway. Symbols * and ** indicated a significant change of p < 0.05 and p < 0.01 respectively.
108
Figure 6.6 Mean change in stance foot CoM position (mm) and relative error (%) with respect to the original foot trajectory pre-to-post kinematic optimisation. Anterior, medial and superior changes are positive.
109
xviii
List of Tables
Table 2.1 ACL injury focused training interventions
27 - 32
Table 2.2 Laboratory-based, biomechanically-focused training interventions
33 - 35
Table 2.3 Field-based, biomechanically-focused training interventions
36 - 37
Table 3.1 Mean knee flexion angle and range of motion (RoM) during the weight acceptance phase of stance for all running tasks. BTT and ST groups across both testing sessions 1 and 2 were pooled together.
57
Table 3.2 Mean peak flexion, valgus and internal rotation (Int. Rot.) knee moments of both training groups across testing session 1 and 2 for all running tasks.
57
Table 3.3 Pearson correlation (R2), 95% confidence interval (95% CI) and limits of agreement (LoA) for change of direction (CoD) angle, pre-contact (PC) velocity and CoD velocity between testing session 1 and 2 for all running tasks.
58
Table 3.4 Mean sidestep CoD angle, CoD velocity and PC velocity for both training groups and across all running tasks. PC velocity was reported for testing sessions 1 and 2.
58
Table 4.1 Muscles grouped according to ability to produce knee moments during flexion, extension, varus, valgus, internal and external rotation degree-of-freedom from 20 to 50 degrees of knee flexion [4, 5, 12, 19].
72
Table 4.2 TMA and DCCR of the muscles crossing the knee with flexion/extension (F/E) and medial/lateral (M/L) moment arms. Data is presented for testing sessions 1 and 2, during both the pre-contact and weight acceptance phases of running and sidestepping. ST and BTT groups were pooled together unless an interaction was observed. DCCR > 0 co-contraction is directed towards muscles with flexion and/or medial moment arms. DCCR < 0 co-contraction is directed towards muscles with extension and/or lateral moment arms. DCCR = 0 maximal co-contraction.
74
Table 4.3 Hamstring-TMA and DCCR of the semimembranosus/biceps femoris (SM/BF) muscles. Data is presented for testing sessions 1 and 2, however the ST and BTT groups as well as the data during the pre-contact and weight acceptance phases of running and sidestepping were pooled.
75
xix
Table 4.4 Mean hip torque, knee torque, CMJ height and full body balance score measures for the ST and BTT test groups between testing sessions 1 and 2.
75
Table 4.5 Relevant TMA and DCCR were calculated during PpSS before and after neuromuscular training from data presented by Zebis et al.[16]. The TFL and MG muscles were not recorded by Zebis et al.[16], so were not used to calculate TMA or the DCCR. It should also be noted that the pre-contact phase in Zebis et al.[16] was 10 ms prior stance foot contact, while in this study it was 50 ms.
77
Table 6.1 Individual simulation (Sim), mean (μ) differences of critical joint coordinates (deg) and mean WB CoM position relative to stance foot CoM position (m) pre-to-post kinematic optimisation. Anterior, medial and superior changes in degrees are positive. Anterior and medial are both towards the desired change of direction pathway. The symbol "--" means the variable was not identified as a critical joint coordinate.
110
xx
List of Tables
Table 2.1 ACL injury focused training interventions
27 - 32
Table 2.2 Laboratory-based, biomechanically-focused training interventions
33 - 35
Table 2.3 Field-based, biomechanically-focused training interventions
36 - 37
Table 3.1 Mean knee flexion angle and range of motion (RoM) during the weight acceptance phase of stance for all running tasks. BTT and ST groups across both testing sessions 1 and 2 were pooled together.
56
Table 3.2 Mean peak flexion, valgus and internal rotation (Int. Rot.) knee moments of both training groups across testing session 1 and 2 for all running tasks.
57
Table 3.3 Pearson correlation (R2), 95% confidence interval (95% CI) and limits of agreement (LoA) for change of direction (CoD) angle, pre-contact (PC) velocity and CoD velocity between testing session 1 and 2 for all running tasks.
57
Table 3.4 Mean sidestep CoD angle, CoD velocity and PC velocity for both training groups and across all running tasks. PC velocity was reported for testing sessions 1 and 2.
58
Table 4.1 Muscles grouped according to ability to produce knee moments during flexion, extension, varus, valgus, internal and external rotation degree-of-freedom from 20 to 50 degrees of knee flexion [4, 5, 12, 19].
71
Table 4.2 TMA and DCCR of the muscles crossing the knee with flexion/extension (F/E) and medial/lateral (M/L) moment arms. Data is presented for testing sessions 1 and 2, during both the pre-contact and weight acceptance phases of running and sidestepping. ST and BTT groups were pooled together unless an interaction was observed. DCCR > 0 co-contraction is directed towards muscles with flexion and/or medial moment arms. DCCR < 0 co-contraction is directed towards muscles with extension and/or lateral moment arms. DCCR = 0 maximal co-contraction.
73
Table 4.3 Hamstring-TMA and DCCR of the semimembranosus/biceps femoris (SM/BF) muscles. Data is presented for testing sessions 1 and 2, however the ST and BTT groups as well as the data during the pre-contact and weight acceptance phases of running and sidestepping were pooled.
74
xxi
Table 4.4 Mean hip torque, knee torque, CMJ height and full body balance score measures for the ST and BTT test groups between testing sessions 1 and 2.
74
Table 4.5 Relevant TMA and DCCR were calculated during PpSS before and after neuromuscular training from data presented by Zebis et al.[16]. The TFL and MG muscles were not recorded by Zebis et al.[16], so were not used to calculate TMA or the DCCR. It should also be noted that the pre-contact phase in Zebis et al.[16] was 10 ms prior stance foot contact, while in this study it was 50 ms.
76
Table 7.1 Individual simulation (Sim), mean (μ) differences of critical joint coordinates (deg) and mean WB CoM position relative to stance foot CoM position (m) pre-to-post kinematic optimisation. Anterior, medial and superior changes in degrees are positive. Anterior and medial are both towards the desired change of direction pathway. The symbol "--" means the variable was not identified as a critical joint coordinate.
107
xxii
Acknowledgements
There are many people that have supported me personally and professionally to allow me to complete and present this dissertation. I would like to thank Dr James J Dowling for providing me with both encouragement and perspective through this winding and sometimes turbulent journey. I would like to thank my supervisors Prof. David Lloyd, Prof. Bruce Elliott and Dr Jeffery Reinbolt. You have taught me that it is through asking the right questions that leads you to the correct answers. I would also like to thank my colleagues:
I would like to acknowledge technical staff at the School of Sport Science, Exercise and Health for both their technical expertise and support during experimental data collections. I would like to thank my participants who gave up their time voluntarily. Without you, research would not be possible. Finally, to my siblings Ryan, Andrew and Niki. To my beautiful nephews Logan and Lucas. To my parents Rick and Judy. To my grandparents Lily, Max, Helen and Pete. You have all provided me with a stable and constant source of support in which to pursue my passions. I thank you all from the bottom of my heart.
Research Assistants
– Dr Alasdair Dempsey
– Dr Tim Doyle
Lab Assistants
– Dr Massimo Sartori*
– Dr Kane Middleton
– Mr Matt Sweeney
– Mr James Dunne
*Visiting Scholar (U Padova)
NMBL
– Prof. Scott Delp
– Dr Ayman Habib
– Mr Sam Hamner
– Mr Matt Demers
– A/Prof. Jeff Reinbolt
– A/Prof. Thor Besier
– Prof. Caroline Finch
xxiii
Statement of Candidate Contribution
The work involved in designing, conducting and analysing the studies described in this
dissertation were primarily performed by Cyril J. Donnelly (candidate). The thesis
outline and experimental design was planned and developed by the candidate, with
consultation from Prof. Bruce Elliott and Prof. David Lloyd (the candidate’s
supervisors). We would like to acknowledge Caroline Finch (Monash University), Dr
Tim Doyle (The University of Western Australia) and Dr Dara Twomey (University of
Ballarat) for assisting with the experimental design and training protocols highlighted in
chapters three and four. We would also like to acknowledge the work of external
supervisor A/Prof. Jeffery Reinbolt for his assistance in the methodological
developments associated with chapters five and six. The final thesis was drafted by
the candidate, with Prof. Bruce Elliott and Prof. David Lloyd providing editorial
feedback.
For each individual chapter there are multiple authors that should also be recognised:
Chapter 2 Publication
1. Donnelly, C.J., Elliott, B.C., Ackland T.R., Doyle T.L.A, Besier T.F., Finch, C.F., Cochrane, J.L., Dempsey A.R., and Lloyd, D.G. (2012). An anterior cruciate ligament injury prevention framework: Incorporating the recent evidence. Res Sports Med. doi:10.1080/15438627.2012.680989.
Conference Proceeding 1. Andrew, N., Gabbe, B., Cook, J., Lloyd, D., Donnelly, C.J., Nash, C.,
Donaldson, A., White, P., Finch., C. What is the evidence-base for exercise as a lower limb injury prevention strategy in community Australian Football? Australian Conference of Science and Medicine in Sport. Fremantle, October 19 – 22, 2011.
Chapter 3 Publication
1. Donnelly C.J., Elliott, B.C., Doyle, T.L.A., Finch, C.F., Dempsey, A.R. and Lloyd, D.G. (2012). Changes in knee joint biomechanics following balance and technique training and a season of Australian football. Br J Sports Med. doi: 10.1136/bjsports-2011-090829.
Conference Proceeding 1. Donnelly, C.J., Doyle, T., Finch, C.F., Elliott, B. and Lloyd, D.G. (2009). The
influence of balance and technique training on knee loading and risk of ACL injury during sidestepping. In Proceedings of The XXII Congress of the International Society of Biomechanics, Cape Town, South Africa, July 5 -9, 2009.
xxiv
Chapter 4 Conference Proceeding
1. Donnelly, C.J., Elliott, B., Doyle, T., Finch, C.F., Dempsey, A. and Lloyd, D.G. Neuromuscular adaptations to balance and technique training during sidestepping: Implications for ACL injury risk. In Proceedings of The Annual Conference of the International Society of Biomechanics in Sport, Porto, Portugal, June 27 – July 1, 2011.
Chapters 5 & 6 Publication
1. Donnelly, C.J., Elliott, B., Lloyd, D.G. and Reinbolt, J.A. (2012). Optimizing Whole body Kinematics to minimize valgus knee loading during sidestepping: Implications for ACL injury risk. J Biomech. 45:1491-1497,
Conference Proceedings 1. Donnelly, C.J., Elliott, B., Lloyd, D.G. and Reinbolt, J.A. Optimizing whole-body
kinematics to minimise valgus knee loading during single-leg landing: Implications for ACL injury risk. In Proceedings of the XXIII Congress of the International Society of Biomechanics, Brussels, Belgium, July 3 -7, 2011.
2. Donnelly, C.J., Elliott, B., Lloyd, D.G. and Reinbolt, J.A. Kinematic adaptations to minimise valgus knee loading during sidestepping: Implications for ACL injury risk. In Proceedings of The 6th World Congress on Biomechanics, Singapore, August 1- 6, 2010.
3. Reinbolt, J.A. & Donnelly, C.J. Improving Computed Muscle Control through Optimization to Generate Dynamic Simulations of Overground-running. In Proceedings of The Eleventh International Symposium on the 3D analysis of Human Movement. San Francisco, California, July 14-16, 2010.
1
CHAPTER 1
INTRODUCTION
1.1 BACKGROUND
Anterior cruciate ligament (ACL) injuries in sport are common (Janssen et al., 2011),
and associated with a high financial and personal cost. It is estimated that 1.3
professional Australian football players (Orchard and Seward, 2009) and 1.15
competitive amateur soccer players (Caraffa et al., 1996) per team per year ruptures
their ACL during play. The average cost of an ACL reconstruction and associated
rehabilitation is approximately 11,157 NZD (Gianotti et al., 2009). New Zealand and
Australia spend approximately 17.4 million NZD (Gianotti et al., 2009) and 75 million
AUD (Janssen et al., 2011) on ACL injuries each year. Extrapolating from figures
reported by Gianotti et al. (2009) and current world population estimates (The World
Bank, June, 2010), the United States spends approximately 1 billion USD on ACL
injury management each year. Of the ACL injuries reported in Cochrane et al. (2007)
(D.G. Lloyd, personal communication, October 20th, 2008), 20% of Australian football
players were not capable of returning to competition one year post injury. Over 50% of
ACL injured athletes were reported as not capable of returning to the same level of
competition two years post reconstruction (Dunn and Spindler, 2010), a percentage
that increases to approximately 70% in three years (Roos et al., 1995). Furthermore,
following an ACL rupture accompanied by a meniscal tear, the probability an athlete
will develop radiographic diagnosed knee osteoarthritis (OA) within 10 to 15 years
increases by 20-50% (Oiestad et al., 2009). Ruptures to the ACL are therefore
considered one of the most costly knee injuries an athlete can sustain in sport.
Over one-half of all ACL injuries occur during non-contact situations (Cochrane et al.,
2007; Koga, et al., 2010), with almost all occurring during either sidestepping or single-
leg landing (Cochrane et al., 2007; Koga et al., 2010). Biomechanical analysis of
sidestepping and single-leg landing have shown that internal rotation and/or valgus
knee moments are elevated (Besier et al., 2001a; Besier et al., 2001b; Cochrane et al.,
2010; McLean et al., 2010); the same loading patterns that elevate ACL strain
measured in cadaveric knee models (Markolf et al., 1995; Withrow et al., 2006). Peak
in-vivo ACL strain also corresponds with peak vertical ground reaction forces during
sport tasks characterized by a rapid deacceleration phase (Cerulli et al., 2003), like
sidestepping (Jindrich et al., 2006). With peak vertical ground reaction forces and
valgus knee moments observed during the weight acceptance (WA) (first 20-30%) of
2
sidestepping (Besier et al., 2001a; Cochrane et al., 2010; Dempsey et al., 2009) and
single-leg landing (McLean et al., 2010), this is when ACL injury risk is thought to be
the greatest.
Ultimately, the mechanism of an ACL injury is that the forces applied to the ligament
are greater than its ability to sustain the load (Lloyd, 2001). Training interventions are
therefore generally focused on protecting the ACL from external joint loading by 1)
changing an athlete’s technique during a sporting task to reduce external joint loading,
and 2) increase the strength and/or the activation of muscles supporting the knee and
ACL when external knee loading is elevated.
Biomechanically-focused training interventions like plyometric, balance, resistance
and/or technique training have shown to be effective in reducing peak knee loading and
increasing medial hamstring activation during landing and sidestepping tasks (Chappell
and Limpisvasti, 2008; Cochrane et al., 2010; Dempsey et al., 2009; Hewett et al.,
1996; Myer et al., 2005; Zebis et al., 2008). These results provide a rationale for the
use of training to reduce ACL injury risk; however, the biomechanical mechanisms by
which these training interventions act are still not well understood. Additionally, the
aforementioned training interventions have all been performed under ‘ideal’ training
settings, meaning it is unknown if these laboratory based finding can be translated into
‘real-world’ community level training environments.
1.2 STATEMENT OF THE PROBLEM
The efficacy of plyometric, balance, resistance and/or technique training in reducing
peak knee loading and/or increasing muscular support have yet to be tested in ‘real-
world’ community level training environments. It is also not well understood how
training protocols like technique training act to influence external knee loading and ACL
injury risk during high risk sporting tasks like sidestepping and single-leg landing. From
both approaches, we will be better able to develop ACL injury prevention training
protocols that target the critical and modifiable risk factors associated with ACL injury
risk. Through this approach we may more effectively transfer positive laboratory-based
training effects to ‘real-world’ training environments and observe reductions of ACL
injury rates across heterogeneous athletic populations in the future.
3
1.3 AIMS AND HYPOTHESES
Chapter 2: An anterior cruciate ligament injury prevention framework: Incorporating the
recent evidence
Aims
Develop an ACL injury prevention framework specific to, and detailed for
intrinsic factors associated with non-contact ACL injuries
Using current empirical evidence, provide a rationale for the design of ACL
injury prevention training protocols, with the goal of reducing ACL injury rates in
the future.
Chapter 3: Changes in knee joint biomechanics following balance and technique
training and a season of Australian football
Aims
Determine if balance and technique training, implemented adjunct to pre-
season and regular season Australian football training is effective in reducing
peak knee moments during the weight acceptance phase of pre-planned and
unplanned sidestepping.
Determine if an Australian football player’s knee joint biomechanics changes
over a season of Australian football.
Hypotheses
Balance and technique training will reduce both peak valgus and internal
rotation knee moments during the weight acceptance phase of anticipated and
unanticipated sidestepping.
Peak valgus and internal rotation knee moments during the weight acceptance
phase of anticipated and unanticipated sidestepping will not change over a
season of Australian football.
Chapter 4: Changes in muscle activation following balance and technique training and
a season of Australian football
Aims
Determine if balance and technique training implemented adjunct to pre-season
and regular season Australian football training influences the activation patterns
4
of the muscles crossing the knee during pre-planned and unplanned
sidestepping.
Determine if an Australian football player’s muscle activation changes over a
normal season of Australian football.
Determine if changes in muscle activation following balance and technique
training are proportional to changes in knee loading during pre-planned and
unplanned sidestepping.
Determine if changes in muscle activation following a season of Australian
football are proportional to changes in knee loading during pre-planned and
unplanned sidestepping.
Hypotheses
Balance and technique training will:
i. Increase the total muscle activation of the muscles crossing the knee
during the pre-contact phase of pre-planned and unplanned
sidestepping.
ii. Increase the co-contraction between knee flexors and extensors during
the pre-contact phases of pre-planned and unplanned sidestepping.
iii. Increase the relative activation of muscles with medial moment arms
during pre-planned sidestepping.
The total activation of the muscles crossing the knee during the pre-contact and
weight acceptance phases of pre-planned and unplanned sidestepping will not
change over a season of Australian football.
The directed co-contraction ratios of the muscles crossing the knee during the
pre-contact and weight acceptance phases of pre-planned and unplanned
sidestepping will not change over a season of Australian football.
Pre-contact total muscle activation following balance and technique training will
be greater than changes in knee loading during the weight acceptance phase of
pre-planned and unplanned sidestepping.
Pre-contact total muscle activation following a season of Australian football will
be similar to changes in knee loading during the weight acceptance phase of
pre-planned and unplanned sidestepping.
5
Chapter 5: An open-source computational method to optimise simulated human motion
to reduce valgus knee loading during sidestepping and single-leg landing.
Aims
Develop a simplified method to create dynamically consistent simulations of
human motion.
Use the open-source musculoskeletal software OpenSim and The Residual
Reduction Algorithm to develop a method to optimise a simulation’s kinematics
to minimise peak valgus knee torques during the weight acceptance phase of
sidestepping
Use the open-source musculoskeletal software OpenSim and The Residual
Reduction Algorithm to develop a method to optimise a simulation’s kinematics
to minimise peak valgus knee torques during the weight acceptance phase of
single-leg landing.
Hypotheses
The Residual Reduction Algorithm in OpenSim with an outer-level optimisation
method can be used to create dynamically consistent simulations of human
motion.
The Residual Reduction Algorithm in OpenSim can be used to identify causal
links between a simulations whole-body kinematics and valgus knee moments
during the weight acceptance phase of unplanned sidestepping.
The Residual Reduction Algorithm in OpenSim can be used to identify causal
links between a simulations whole-body kinematics and valgus knee moments
during the weight acceptance phase of single-leg landing.
Chapter 6: Optimizing whole-body kinematics to minimise valgus knee loading during
sidestepping: implications for ACL injury risk.
Aims
Use the open-source musculoskeletal modelling platform OpenSim, an outer-
level optimisation technique and the Residual Reduction Algorithm to identify a
generalised kinematic strategy to reduce peak valgus knee moments during the
weight acceptance phase of unplanned sidestepping.
Hypotheses
Frontal plane upper body kinematics will be related to increased peak valgus
knee moments during the weight acceptance phase of unplanned sidestepping.
6
Multiple kinematic changes along the kinematic chain will be used to minimising
peak valgus knee moments during the weight acceptance phase of unplanned
sidestepping.
1.4 LIMITATIONS
It is assumed the sample is representative of an amateur level, community
based athletic population.
It is assumed that athletes use the same sidestepping and landing techniques
during testing (in a laboratory) as they would display in a sporting situation
(during a game or training).
It was assumed a computer monitor was an ecologically valid signal to initiate
an unanticipated sidestepping condition.
Taking off and landing with the same leg during the laboratory based single-leg
landing tasks is representative of techniques used by athletes in game
situations.
Increasing the tracking of the kinematic markers on the foot during an
optimisation is similar to using a foot contact model.
1.5 DELIMITATIONS
Thirty-four athletes, pre-to-post training or a sub-set of this 34 was used for all
analysis.
All participants ran and conducted sidestepping tasks at velocity between 4.5
ms-1 and 5.5 ms-1 during testing.
A change of direction angle of 45° is representative of the motion of a sidestep.
All participants were instructed to take off and land with the same leg during
dingle-leg landing tasks.
During unanticipated sidestepping, all participants were signalled to change
direction when they were approximately 1.5 m from the force plate or
contralateral leg toe off.
7
1.6 DEFINITION OF TERMS
Centre of Mass (CoM) The average location of a system’s total mass or a point where all of the mass of a system is concentrated.
Centre of Pressure (CoP) Location on a force platform where the total sum of external forces acts upon a system.
Contralateral On the opposite side relative to a reference structure.
Co-contraction The simultaneous contraction of agonist and antagonist muscles around a given joint.
Cross-over step During stance, the whole-body CoM is directed laterally towards the support limb, while the swing leg is moved across the upper body midline and the support limb.
Degree of freedom An independent set of allowable displacements and/or rotations between two bodies defining or describing a joint’s motion.
‘Dynamic valgus’
Described as the dynamic motion of the knee joint in the frontal plane moving into valgus posture. This motion is generally observed during the weight acceptance phase of landing.
Epidemiology
The study of health-event, health characteristic, or health-determinant patterns in a population.
Force The concept of force is used to describe an influence which causes a free body to undergo an acceleration or which can cause a flexible/compliant object to deform (e.g. bone, cartilage, ligaments and tendons).
Inverse Dynamics A method for computing forces and/or moments of force (torques) based on the kinematics (motion) of a body and their inertial properties (mass, CoM position and moment of inertia).
Inverse Kinematics A global optimisation method (weighted least-squares) used to calculate a skeletal model’s generalised coordinates (i.e. q or joint angles). This is done by minimising the squared distances between the rigid segment markers of the skeletal model and the experimentally recorded kinematics by adjusting the skeletal model’s generalised coordinates (q).
Ipsilateral On the same side relative to a reference structure.
In-silico Experiments performed using a computer or though computer simulation.
In-vivo An experimental design that uses a whole, living organism as opposed to a partial or dead organism.
8
Kinematics The branch of mechanics that studies the motion of a body or a system of bodies without consideration given to its mass or the forces acting on it.
Kinetics
The branch of mechanics that studies the relationships between the forces and torques causing the motion of a body or system.
Least Mean Squared Error Data fitting approach designed to approximate the solution of over-determined systems (i.e. more equations than unknowns). Least-squares minimises the sum of squared residuals, providing a solution with minimised difference between an observed value and a model.
Moment (Also known as: Torque or moment of force)
It is the tendency of a force to rotate an object about an axis. Loosely speaking, torque is a measure of the turning force on an object such as a bolt or a flywheel.
OpenSim Open-source physics based musculoskeletal modelling software developed at Stanford University in 2006 to answer clinical based biomechanical questions.
Optimised/Optimisation In mathematics and computer science, optimisation refers to choosing the best element or set of elements from some larger set of available alternatives. In the simplest case, this means solving problems in which one seeks to minimise or maximise a real function by systematically choosing the values of real or integer variables from within an allowable set of alternatives.
Osteoarthritis (Also known as: OA, degenerative arthritis, degenerative joint disease)
It is a group of diseases and mechanical abnormalities involving degradation of joints, including articular cartilage and the subchondral bone next to it.
Pre-contact A phase of motion 50 ms prior to weight acceptance.
Residual Force/Moment Forces and moments not solved during inverse dynamics. These represent the errors and assumptions in the modelling process (i.e. joint centre and inertial estimates). In OpenSim, a 6 DoF joint between the pelvis and ground is used to hold these forces and moments, satisfying Newton’s second law (∑Fmodel + ∑Fresiduals = GRF).
Residual Reduction Algorithm (RRA)
Produces a set of actuator forces (i.e. joint torques) to generate joint motions that track a desired set of generalised coordinates, while minimising the model’s residual forces and moments (i.e. modelling errors). The result is simulation that tracks the experimentally recorded GRF with dynamic consistency.
Sidestep During stance, the whole-body CoM is directed laterally away from the support limb, while the swing leg is moved away from the upper body midline and the support limb.
9
Strain The relative change in length of a tissue in response to an external force per unit area or stress.
Weight Acceptance
A phase of motion within the stance phase of gait, from heel strike to the first trough in the vertical ground reaction force vector. This usually occurs within the first 20-30% of stance.
Valgus moment (Also known as: Abduction)
The distal end of the shank segment is forced laterally causing an abduction moment at the knee.
Varus moment (Also known as: Adduction)
The distal end of the shank segment is forced medially causing an adduction moment at the knee.
10
1.7 LIST OF ABREVIATIONS
3D Three-dimensional
ACL Anterior cruciate ligament
A/D Analogue to digital
AF Australian football
Ag/AgCl Silver/Silver chloride
AMB Anteromedial bundle
ANOVA Analysis of variance
A/P Anterior/Posterior
AUD Australian dollar
BF Biceps femoris
BTT Balance and technique training
CMJ Countermovement jump
CMR Common-mode rejection ratio
CoD Change of direction
CoM Centre of mass
CoP Centre of pressure
DCCR Directed co-contraction ratio
DLL Double-leg landing
DoF Degree of freedom
F Female
F/E Flexion/Extension
GRF Ground reaction force
I/E Internal/external
ID Inverse dynamics
IK Inverse kinematics
I/S Inferior/Superior
LG Lateral gastrocnemius
LoA Limit of agreement
M Male
MG Medial gastrocnemius
M/L Medial/Lateral
NZD New Zealand dollar
OA Osteoarthritis
PAFIX Preventing Australian football Injuries through eXercise
PC Pre-contact
PLB Posterolateral bundle
11
PpSS Pre-planned sidestep
RCT Randomized control trial
RF Rectus femoris
RoM Range of motion
RRA Residual Reduction Algorithm
sEMG Surface electromyography
SLL Single-leg landing
SM Semimembranosus
SM/BF Semimembranosus/Biceps femoris
ST Sham training
TFL Tensor fasciae latae
TMA Total muscle activation
TRIPP Translating Research Into injury Prevention Practice
UnSS Unplanned sidestep
US United States
USD United States dollar
UWA University of Western Australia
VL Vastus lateralis
VM Vastus medialis
V/V Varus/Valgus
WB Whole-body
WBB Whole-body balance
WA Weight acceptance
WAAFL Western Australian Amateur Football League
12
Reference list chapter 1
Besier, T.F., Lloyd, D.G., Ackland, T.R., Cochrane, J.L., 2001a. Anticipatory effects on knee joint loading during running and cutting maneuvers. Med Sci Sports Exerc. 33 (7), 1176-1181. Besier, T.F., Lloyd, D.G., Cochrane, J.L., Ackland, T.R., 2001b. External loading of the knee joint during running and cutting maneuvers. Med Sci Sports Exerc. 33 (7), 1168-1175. Caraffa, A., Cerulli, G., Projetti, M., Aisa, G., Rizzo, A., 1996. Prevention of anterior cruciate ligament injuries in soccer. A prospective controlled study of proprioceptive training. Knee Surg Sports Traumatol Arthrosc. 4 (1), 19-21. Cerulli, G., Benoit, D.L., Lamontagne, M., Caraffa, A., Liti, A., 2003. In vivo anterior cruciate ligament strain behaviour during a rapid deceleration movement: Case report. Knee Surg Sports Traumatol Arthrosc. 11 (5), 307-311. Chappell, J.D., Limpisvasti, O., 2008. Effect of a neuromuscular training program on the kinetics and kinematics of jumping tasks. Am J Sports Med. 36 (6), 1081-1086. Cochrane, J.L., Lloyd, D.G., Besier, T.F., Elliott, B.C., Doyle, T.L., Ackland, T.R., 2010. Training affects knee kinematics and kinetics in cutting maneuvers in sport. Med Sci Sports Exerc. 42 (8), 1535-1544. Cochrane, J.L., Lloyd, D.G., Buttfield, A., Seward, H., McGivern, J., 2007. Characteristics of anterior cruciate ligament injuries in australian football. J Sci Med Sport. 10 (2), 96-104. Dempsey, A.R., Lloyd, D.G., Elliott, B.C., Steele, J.R., Munro, B.J., 2009. Changing sidestep cutting technique reduces knee valgus loading. Am J Sports Med. 37 (11), 2194-2200. Dunn, W.R., Spindler, K.P., 2010. Predictors of activity level 2 years after anterior cruciate ligament reconstruction (aclr): A multicentre orthopaedic outcomes network (moon) aclr cohort study. Am J Sports Med. 38 (10), 2040-2050. Gianotti, S.M., Marshall, S.W., Hume, P.A., Bunt, L., 2009. Incidence of anterior cruciate ligament injury and other knee ligament injuries: A national population-based study. J Sci Med Sport. 12 (6), 622-627. Hewett, T.E., Stroupe, A.L., Nance, T.A., Noyes, F.R., 1996. Plyometric training in female athletes. Decreased impact forces and increased hamstring torques. Am J Sports Med. 24 (6), 765-773. Janssen, K.W., Orchard, J.W., Driscoll, T.R., van Mechelen, W., 2011. High incidence and costs for anterior cruciate ligament reconstructions performed in australia from 2003-2004 to 2007-2008: Time for an anterior cruciate ligament register by scandinavian model? Scand J Med Sci Sports. doi: 10.1111/j.1600-0838.2010.01253.x Jindrich, D.L., Besier, T.F., Lloyd, D.G., 2006. A hypothesis for the function of braking forces during running turns. J Biomech. 39 (9), 1611-1620. Koga, H., Nakamae, A., Shima, Y., Iwasa, J., Myklebust, G., Engebretsen, L., Bahr, R., Krosshaug, T., 2010. Mechanisms for noncontact anterior cruciate ligament injuries:
13
Knee joint kinematics in 10 injury situations from female team handball and basketball. Am J Sports Med. 38 (11), 2218-2225. Lloyd, D.G., 2001. Rationale for training programs to reduce anterior cruciate ligament injuries in australian football. J Orthop Sports Phys Ther. 31 (11), 645-654; discussion 661. Markolf, K.L., Burchfield, D.M., Shapiro, M.M., Shepard, M.F., Finerman, G.A., Slauterbeck, J.L., 1995. Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res. 13 (6), 930-935. McLean, S.G., Borotikar, B., Lucey, S.M., 2010. Lower limb muscle pre-motor time measures during a choice reaction task associate with knee abduction loads during dynamic single leg landings. Clin Biomech (Bristol, Avon). 25 (6), 563-569. Myer, G.D., Ford, K.R., Palumbo, J.P., Hewett, T.E., 2005. Neuromuscular training improves performance and lower-extremity biomechanics in female athletes. J Strength Cond Res. 19 (1), 51-60. Orchard, J., & Seward, H. (2009). 17th Annual AFL injury Report: 2008. 2010, 1-14. Retrieved from http://www.afl.com.au website: http://www.afl.com.au Oiestad, B.E., Engebretsen, L., Storheim, K., Risberg, M.A., 2009. Knee osteoarthritis after anterior cruciate ligament injury: A systematic review. Am J Sports Med. 37 (7), 1434-1443. Roos, H., Ornell, M., Gardsell, P., Lohmander, L.S., Lindstrand, A., 1995. Soccer after anterior cruciate ligament injury--an incompatible combination? A national survey of incidence and risk factors and a 7-year follow-up of 310 players. Acta Orthop Scand. 66 (2), 107-112. The World Bank Group [Internet]. Washington, DC (USA): World Population Estimates; [cited 2010 June 7]. Available from: http://data.worldbank.org. Withrow, T.J., Hutson, L.J., Wojtys, E.M., Ashton-Miller, J.A., 2006. The effect of an impulsive knee valgus moment on in vitro relative ACL strain during a simulated jump landing. Clin Biomech (Bristol, Avon). 21 (9), 977-83. Zebis, M.K., Bencke, J., Andersen, L.L., Dossing, S., Alkjaer, T., Magnusson, S.P., Kjaer, M., Aagaard, P., 2008. The effects of neuromuscular training on knee joint motor control during sidecutting in female elite soccer and handball players. Clin J Sport Med. 18 (4), 329-337.
14
CHAPTER 2
AN ANTRIOR CRUSCIATE LIGAMENT INJURY PREVENTION FRAMEWORK:
INCORPORATING THE RECENT EVIDENCE
A version of the presented literature review has been accepted for publication in the
Journal of Research in Sports Medicine.
Donnelly, C.J., Elliott, B.C., Ackland T.R., Doyle T.L.A, Besier T.F., Finch, C.F., Cochrane, J.L., Dempsey A.R., and Lloyd, D.G. (2012). An anterior cruciate ligament injury prevention framework: Incorporating the recent evidence. Res Sports Med. doi:10.1080/15438627.2012.680989.
The PhD candidate, Cyril J. Donnelly accounted for 80% of the intellectual property
associated with the final manuscript. Collectively, the remaining authors contributed
20%.
Conflict of Interest: There were no financial or personal relationships with other people
or organizations that could have biased the presented work incorporating
15
Abstract
Anterior cruciate ligament (ACL) injury rates have increased by ∼50% over the last 10
years. These figures suggest that ACL focused research has not been effective in
reducing injury rates among community level athletes. Training protocols designed to
reduce ACL injury rates have been both effective (n = 3) and ineffective (n = 7).
Although a rationale for the use of exercise to reduce ACL injuries is established, the
mechanisms by which they act are relatively unknown. This article provides an injury
prevention framework specific to noncontact ACL injuries and the design of
prophylactic training protocols. It is also apparent that feedback within this framework is
needed to determine how biomechanically relevant risk factors like peak joint loading
and muscular support are influenced following training. It is by identifying these links
that more effective ACL injury prevention training programs can be developed, and, in
turn, lead to reduced ACL injury rates in the future.
Keywords: Injury Prevention; Sport Injuries; Prophylactic; Model
2.1 INTRODUCTION
Anterior cruciate ligament (ACL) ruptures are severe sport injuries, dramatically
affecting an athlete’s ability to return to play following reconstruction (Dunn & Spindler,
2010; Roos, Ornell, Gardsell, Lohmander, & Lindstrand, 1995). Furthermore, following
an ACL rupture, when accompanied by a meniscal injury, the probability that an athlete
will develop radiographic diagnosed knee osteoarthritis (OA) within 10 to 15 years
increases by 20–50% (Oiestad, Engebretsen, Storheim, & Risberg, 2009).
In the United States, ACL injury estimates prior to 1998 were 23/100,000 people per
year, increasing to 35/100,000 people per year in 2006 (Lyman et al., 2009;
TheWorldBank, 2010). These U.S. figures are consistent with current estimates from
both New Zealand (2000–2005) and Scandinavia (2004–2007), which have reported
ACL injury rates of 32–37/100,000 (Gianotti, Marshall, Hume, & Bunt, 2009) and
38/100,000 (Granan, Forssblad, Lind, & Engebretsen, 2009) people per year,
respectively. In Australia (2003–2008), ACL injury rates have been reported to be as
high as 52/100,000 people per year (Janssen, Orchard, Driscoll, & van Mechelen,
2011). Improved injury surveillance, increases in sport participation and exposure, or
rule changes within a sport to increase the speed of play may all have contributed to
the observed increases in ACL injury estimates. However, with such a large increase
16
(∼ 50%) over such a short period of time (∼ 10 years), it is apparent that in the context
of the community level athlete, ACL injury prevention research is not being effectively
translated into injury prevention practice.
Training interventions designed to reduce ACL injury rates in general athletic
populations have been shown to be both effective (Caraffa, Cerulli, Projetti, Aisa, &
Rizzo, 1996; Hewett, Lindenfeld, Riccobene, & Noyes, 1999; Mandelbaum et al.,
2005), and ineffective (Heidt, Sweeterman, Carlonas, Traub, & Tekulve, 2000; Junge,
Rosch, Peterson, Graf-Baumann, & Dvorak, 2002; Myklebust et al., 2003; Pfeiffer,
Shea, Roberts, Grandstrand, & Bond, 2006; Soderman, Werner, Pietila, Engstrom, &
Alfredson, 2000; Steffen, Myklebust, Olsen, Holme, & Bahr, 2008; Wedderkopp,
Kaltoft, Holm, & Froberg, 2003). Although empirical research has shown that balance,
plyometric, and/or neuromuscular training can be used to reduce ACL injury rates, the
mechanisms by which such training protocols act is still relatively unknown, which is
evident from the large number of ACL injury prevention studies with inconclusive
findings.
Ultimately, the mechanism of an ACL injury is that the loads applied to the ligament are
greater than its ability to sustain the load (Lloyd, 2001). All ACL injury prevention
programs, whether designed for males or females, should therefore focus on reducing
the loads applied to the joint and in turn ACL during sporting tasks. The loads applied
to the ACL are influenced by externally applied joint loads, the activation of muscles
capable of supporting these loads, the orientation of the tibiofemoral joint when loads
are applied, as well as the anatomical alignment of the ligament. The focus of this
review is on interventions designed to reduce external joint loads and/or improve
muscular support during noncontact sporting tasks.
Training interventions must be designed to target the causal factors associated with
ACL injury (Lloyd, 2001) if positive treatment effects can be effectively translated to,
and adopted by, community level athletes (Finch, 2006). It is beyond the scope of this
article to describe the epidemiology of ACL injuries and the evidence for their risk
factors in detail. Rather, this article will present a framework for translating ACL
focused research into injury prevention practice in the context of the community level
athlete. Through the development of this framework, a scientific rationale for the design
of ACL injury prevention training protocols will also be presented.
17
2.2 ACL INJURY PREVENTION FRAMEWORK
The six stage injury prevention model to Translate Research into Injury Prevention
Practice (the TRIPP model) proposed by Finch (2006) provides a blueprint for
preventing injuries in sport. Borrowing from the TRIPP model, this article provides an
ACL injury prevention framework specific to, and detailed for, the intrinsic factors
associated with noncontact ACL injuries and an empirically based rationale for the
design of ACL injury prevention training protocols.
2.3 INJURY SURVEILLANCE
General population estimates show 32–52/100,000 people per year rupture their ACL,
with the majority occurring during sport (Gianotti et al., 2009; Granan et al., 2009;
Lyman et al., 2009; Janssen et al., 2011). Retrospective surveys (Gianotti et al., 2009;
Rochcongar, Laboute, Jan, & Carling, 2009) and video analyses of athletes rupturing
their ACL (Cochrane, Lloyd, Buttfield, Seward, & McGivern, 2007; Krosshaug et al.,
2007) show that approximately half occur during noncontact situations. Of these
noncontact injuries, almost all occur during landing or sidestepping, immediately after
foot contact, with the knee near full extension (Cochrane et al., 2007; Koga et al., 2010;
Krosshaug et al., 2007). Further classification of noncontact landing injuries shows that
the majority occur during single-leg landing situations (Cochrane et al., 2007; Koga et
al., 2010)..
2.4 MECHANICAL AETIOLOGY OF ACL INJURY
The ultimate mechanism of an ACL injury is that the forces applied to the ligament are
greater than its ability to sustain the load (Lloyd, 2001). Experimental laboratory, in-
vivo/cadaveric and in-silico research have provided valuable information to better
understand what loading patterns, joint kinematics, and phases of movement are
associated with increased ACL injury risk. Using this information, a model for the
aetiology of ACL injuries can be formulated and, in turn, appropriate countermeasures
developed.
Valgus, internal rotation knee moments and anterior tibial translations relative to the
femur (anterior drawer) all elevate ACL strain in cadaveric knee models (Markolf et al.,
1995; Shin, Chaudhari, & Andriacchi, 2011). However, it is the combined loading of
these moments/forces that contributes to the largest ACL strain measurements and
injury risk. For example, tibiofemoral compression and internal rotation moments
18
(Meyer & Haut, 2008), valgus and internal rotation moments (Shin et al., 2011), and
anterior drawer combined with either valgus or internal rotation moments (Markolf et
al., 1995) all elevate ACL strain more than anterior drawer alone. Simulation studies
(in-silico) support the view that anterior draw alone is not the likely mechanism of ACL
injury; the addition of valgus knee moments are required to achieve injurious loads
(McLean, Huang, Su, & Van Den Bogert, 2004; McLean, Huang, & van den Bogert,
2008).
Laboratory-based analyses of noncontact sidestepping have shown that compared with
straight-line running, peak extension knee moments are similar, while internal rotation
and/or valgus knee moments are up to two-times higher (Besier, Lloyd, Ackland, &
Cochrane, 2001; Besier, Lloyd, Cochrane, & Ackland, 2001; Dempsey et al., 2007).
Valgus and internal rotation knee moments are also elevated during single-leg landing
(McLean, Borotikar, & Lucey, 2010; McLean & Samorezov, 2009). Hewett et al. (2005)
showed valgus knee moments observed during double-leg landing can predict the ACL
injury status of adolescent females with 73% specificity and 78% sensitivity. Again,
these are the same loading patterns shown to elevate ACL strain in cadaveric knee
models (Markolf et al., 1995; Shin et al., 2011). It should be noted that peak internal
rotation and/or valgus knee moments are elevated further when sidestepping (Besier,
Lloyd, Ackland, et al., 2001) and single-leg landing (McLean et al., 2010) are
performed in unplanned situations.
Peak in-vivo ACL strain measured in a single healthy male has been shown to occur
during the weight acceptance (WA) phase of stance (first 20% – 30%) during the
deacceleration phase of a landing task (Cerulli, Benoit, Lamontagne, Caraffa, & Liti,
2003)(Figure 2.1), which are similar to the accelerations observed during the WA
phase of sidestepping (Jindrich, Besier, & Lloyd, 2006). Additionally, WA is where peak
internal rotation and/or external valgus knee moments are observed during
sidestepping (Besier, Lloyd, Ackland, et al., 2001; Besier, Lloyd, Cochrane, et al., 2001;
Cochrane et al., 2010; Dempsey, Lloyd, Elliott, Steele, & Munro, 2009; Dempsey et al.,
2007) and single-leg landing (McLean et al., 2010). It is therefore logical to identify the
WA phase of landing and sidestepping as when the ACL is at greatest risk of injury.
Knee valgus angle or “dynamic valgus” angle during double-leg landing has been
shown to be significantly greater in ACL injured versus uninjured adolescent female
populations and a predictor of ACL injury (R2 = 0.88; Hewett et al., 2005). It should be
appreciated however, that knee range of motion in the varus/valgus degree of freedom
is limited and unlikely to reach a spread of 30° across participants as reported
19
previously (Hewett et al., 2005). Measurements of “dynamic valgus” angles are, to a
certain extent, projections resulting from a combination of femoral internal rotation and
knee flexion, which is likely the reason that the reliability of knee varus/valgus (median
CMC = 0.74) joint angle measurements are substantially lower than knee
flexion/extension (median CMC = 0.96) joint angle measurements (McGinley, Baker,
Wolfe, & Morris, 2009). It is acknowledged that “dynamic valgus” knee postures are
indeed associated with ACL injury risk (Hewett et al., 2005). However, the means by
which athletes attain these postures is likely due to poor hip neuromuscular control
during WA, which has been shown to be associated with peak frontal, sagittal, and/or
transverse plane knee loading during both sidestepping (McLean, Huang, & van den
Bogert, 2005) and single-leg landing (Kipp, McLean, & Palmieri-Smith, 2011).
Figure 2.1
Knee flexion angle is another factor that affects the transfer of external knee loads to
the ACL (Fukuda et al., 2003; Markolf et al., 1995; Wu, Seon, et al., 2010). The ACL
consists of two bundles, the anteromedial bundle (AMB) and posterolateral bundle
(PLB), named from their insertions on the tibial plateau. Direct strain measures of the
AMB and PLB in a cadaveric knee model have shown these bundles function in a
reciprocal manner, with the PLB taut in extension (0°–15°) and the AMB taut in flexion
Anterior cruciate ligament (ACL) strain of the left leg and vertical ground reaction force (GRF) recorded during the stance phase of a stop-landing sport task. Data was then ensemble average of three stop-landing sport tasks. Foot strike occurs at approximately 23% of cycle. The right foot is placed on the ground at approximately 98% and marks the end of the hop cycle. [Adapted from Cerulli et al. (2003)].
20
(60°– 90°)(Gabriel, Wong, Woo, Yagi, & Debski, 2004). However, when quadriceps
muscle forces are simulated, the functioning of the AMB and PLB change and begin
working in a complementary manner, with the peak strain of both bundles observed
near full extension (i.e., 0 and 15 degrees of knee flexion)(Wu, Seon, et al., 2010).
Modelling of the AMB and PLB during the stance phase of gait (0.7 m/s) has shown the
kinematics of the ACL change as a function of knee flexion angle, with peak elongation
observed near full extension (Wu, Hosseini, et al., 2010)(Figure 2.2). These results
support a mechanistic rationale as to why most noncontact ACL injuries occur with the
knee close to full extension (Cochrane et al., 2007; Koga et al., 2010; Krosshaug et al.,
2007).
Figure 2.2
It is clear that increasing knee flexion can reduce ACL strain and hip neuromuscular
control is associated with peak knee loading and ACL injury risk. The role of the
muscles in supporting the hip and knee during sporting tasks, however, should not be
overlooked. As the knee flexes, the moment arms of the muscles crossing the knee
joint change, altering their ability to support external knee loads (Lloyd & Buchanan,
2001). When the knee is flexed, the hamstring muscles are aligned to resist anterior
drawer, while the lateral hamstrings are capable of better supporting internal rotation
moments (Buford, Ivey, Nakamura, Patterson, & Nguyen, 2001). Conversely, the
medial hamstrings and quadriceps both become less capable of supporting valgus
knee moments (Lloyd & Buchanan, 2001). When the knee is extended, the opposite is
true. Further research is needed to determine how knee kinematics and the hip and
trunk musculature influence knee loading during sporting tasks. With more
sophisticated modelling techniques future research may be able to develop subject-
specific models capable of quantifying the complex interaction between hip kinematics,
knee flexion angle, muscle force estimates, knee joint loading, and ultimately ACL
strain to allow for a better assessment of how muscles function to support the knee and
mitigate ACL strain and injury risk during sporting tasks.
Relationship between relative elongation of the anteromedial bundle (AMB) left A and posterolateral bundle (PLB) right B relative to knee flexion angle during stance phase of gait. [Adapted from Wu, Hosseini, et al. (2010)].
21
2.5 COUNTERMEASURES
One logical method to reduce the risk of ACL injury would be to strengthen the tissue
itself, making it more capable of withstanding larger loads. Following a period of
immobilization, mobilization (i.e., exercise) can be used to stimulate collagen
regeneration in rabbit medial collateral ligamentous tissues and Rhesus monkey ACL
tissues to 79% (Woo et al., 1987) and 91% (Noyes, 1977) of the strength of
comparable healthy tissues. Surprisingly, to our knowledge, no published peer-review
study has shown that training can be used to promote collagen regeneration that leads
to significant strength increases in healthy ACL tissues. In fact, research has shown
that post maturation, collagen concentration and ligament force tolerance in healthy
ACL tissues significantly decrease with age (Amiel, Kuiper, Wallace, Harwood, &
VandeBerg, 1991; Noyes & Grood, 1976). This provides a rationale to focus on
reducing the loads applied to the ACL. Two approaches can be used to reduce
ligament loading: (1) change an athlete’s technique during a sporting task to reduce
external joint loading and (2) increase the strength and/or activation of the muscles
supporting the knee and ACL when external joint loading is elevated.
2.6 COUNTERMEASURES: TECHNIQUE AND KNEE LOADING
The potential for upper body segments to influence the loading of distal joints in the
kinematic chain is substantial. Over one-half of a person’s mass is located in the head,
arms, and trunk, which are located over one-half of an individual’s total height from the
ground (Winter, 2005). Hip neuromuscular control (McLean et al., 2005), lateral trunk
flexion (Dempsey et al., 2007), and restraining an athlete’s arm close to midline
(Chaudhari, Hearn, & Andriacchi, 2005) have all been shown to increase valgus and/or
internal rotation knee moments during sidestepping. Hip neuromuscular control has
also been shown to be the primary predictor of both frontal and transverse plane knee
loading during single-leg landing (Kipp et al., 2011).
Altering a person’s technique during sidestepping has been proven effective in
reducing valgus knee moments during sidestepping (Dempsey et al., 2009). The three
recommendations made to athletes were to place their stance foot closer to the body’s
midline, while keeping their torso upright and rotated toward the desired direction of
travel (Dempsey et al., 2009). Motor control strategies to reduce external knee loading
during single-leg landing tasks have not yet been tested.
22
Identifying direct, causative links between an athlete’s kinematics and joint loading is
difficult when treating sidestepping and single-leg landing as multisegment, dynamic
movements. As such, limited causal information linking critical aspects of the
movement pattern to the knee loading patterns associated with elevated ACL injury risk
is available. Future research is needed to establish these causal links if more refined
and effective ACL injury prevention training protocols can be developed in the future.
2.7 COUNTERMEASURES: NEUROMUSCULAR SUPPORT
There is no single muscle crossing the knee capable of simultaneously supporting the
knee from externally applied flexion, valgus, and internal rotation knee moments. For
this reason, multiple muscle activation strategies can be used to reduce ACL injury risk
during sidestepping and single-leg landing.
When simulating the contact phase of landing in a cadaveric knee model, Hashemi et
al. (2010) found that increased quadriceps force in the pre-contact (PC) phase of
landing resulted in lower ACL strain during the impact phase (Figure 2.3). The
reductions in ACL strain were attributed to the quadriceps’ ability to prevent the tibia
from translating relative to the femur by both increasing joint stiffness at low knee
flexion angles and producing posteriorly directed joint reaction forces past 20◦ of knee
flexion (Hashemi et al., 2010). These results are supported by Wu, Seon, et al. (2010)
who has shown that the application of a 400 N quadriceps force can reduce peak AMB
force by almost 50% (123N to 75N).
Due to their line of action, hamstring muscle force can reduce ACL tension from 15° to
45° of knee flexion (More et al., 1993). Anterior cruciate ligament (ACL) strain is
reduced further however, when the hamstrings are co-contracted with the quadriceps
(Withrow, Huston, Wojtys, & Ashton-Miller, 2008). The co-contraction of the quadriceps
and hamstring muscle groups reduces ACL tension from 15°– 60° of knee flexion by
resisting the displacement of the tibia relative to the femur in all three planes of motion
(Li et al., 1999; Withrow et al., 2008).
Valgus and internal rotation knee moments can be supported with the activation of
specified muscles crossing the knee joint (Lloyd, Buchanan, & Besier, 2005).
Generally, medial knee muscles have moment arms capable of supporting valgus knee
moments (Buchanan & Lloyd, 1997; Lloyd 270 & Buchanan, 1996, 2001; Lloyd et al.,
2005) and considered an appropriate neuromuscular strategy for supporting the knee
23
and ACL from external valgus knee moments (Buchanan & Lloyd, 1997; Lloyd &
Buchanan, 2001).
Figure 2.3
In summary, appropriate muscle activation strategies to counter applied flexion, valgus,
and/or internal rotation knee moments during sidestepping and single-leg landing
include generalized hamstring/quadriceps co-contraction, superimposed with the
increased activation of muscles with flexion and/or medial moment arms.
2.8 ACL FOCUSED TRAINING INTERVENTION PROTOCOLS IN SPORT
A review of the current ACL injury prevention literature was conducted. The databases
of Science Direct, EMBASE, Web of Science, AUSport Med, Medline, and Ovid SP
were searched. The search was restricted to human studies, conducted between 1990
and July 2011, and written in English. Search terms included (sprain∗ or injur∗ or
rupture∗ or strain∗ or tear∗ or trauma∗ or pain∗ or stiff∗) AND (prevent∗ or risk∗ or rate∗
or safe∗ or prophylactic∗) AND (tibiofemoral∗ or knee∗ or ACL or anterior cruciate∗ or
cruciate∗). From the six databases, 2,541 titles and abstracts were assessed and
reviewed. Of these, 53 manuscripts were considered further, and then 20 manuscripts
were selected for inclusion in the final review. Inclusion criteria were restricted to
Linear trend line for landing ACL strain versus quadriceps pre-activation forces for pool of all knees (peak strains) measured during upward impulse. (Mean ± standard error of the mean). [Adapted from Hashemi et al. (2010)].
24
training interventions measuring changes in ACL injury rates (n = 10), laboratory-
based, biomechanically focused training interventions (n = 6) and field-based,
biomechanically focused training interventions (n = 4).
For the purpose of this review, when appropriate, ACL injury prevention protocols were
classified into four general categories: (1) plyometric training: exercises with ballistic
movements containing both concentric and eccentric phases (i.e., jumping and
landing); (2) balance training: postural exercises with an unstable base of support
and/or single-leg support with or without external perturbations; (3) technique training:
instructional feedback immediately prior to, following, or during a sport task (i.e.,
running, landing, and sidestepping); and (4) resistance training: movements performed
against external forces progressively overloading isolated muscle groups.
Following a review of the literature, it was found that combinations of plyometric,
balance, resistance, and/or technique training can be used to reduce ACL injury rates
in athletic populations (Table 2.1) (Caraffa et al., 1996; Hewett et al., 1999;
Mandelbaum et al., 2005). Similar ACL injury prevention protocols, however, have
been shown to be inconclusive or ineffective in reducing ACL injury rates in general
athletic populations (Heidt et al., 2000; Junge et al., 2002; Myklebust et al., 2003;
Pfeiffer et al., 2006; Soderman et al., 2000; Steffen et al., 2008; Wedderkopp et al.,
2003). Athlete compliance with the training protocols (Myklebust et al., 2003;
Soderman et al., 2000; Steffen et al., 2008), intervention exposure (Soderman et al.,
2000; Steffen et al., 2008), and/or specificity of the training intervention (Junge et al.,
2002; Myklebust et al., 2003) were identified as some external factors that may have
prevented the combined effects of plyometric, balance, resistance, and/or technique
training from being translated into reductions in ACL injury rates. Nonetheless, these
conflicting results illustrate the point that biomechanical measurements like technique,
joint loading, and muscle support during sidestepping and single-leg landing need to be
measured in parallel with changes in ACL injury rates. Through this approach, it may
be possible to identify the biomechanical mechanisms by which training influences the
factors associated with noncontact ACL injuries and why a particular training protocol
led to a positive or inconclusive training outcome.
Laboratory-based, biomechanically focused training interventions (Table 2.2) such as
plyometric, balance, resistance, and/or technique training have been proven effective in
reducing peak valgus knee moments (Hewett, Stroupe, Nance, & Noyes, 1996; Myer,
Ford, Palumbo, & Hewett, 2005) and increasing knee flexion angle (Myer, Ford,
McLean, & Hewett, 2006; Myer et al., 2005) during pre-planned double-leg (Hewett et
al., 1996; Myer et al., 2006; Myer et al., 2005) and single-leg (Myer et al., 2006) landing
25
tasks. These results demonstrate that training as a whole is effective in altering an
athlete’s knee joint biomechanics and subsequent ACL injury risk. Still, it is unclear
what the most appropriate training protocols or combinations of exercises should be for
reducing ACL injury risk during double-leg or single-leg landing tasks.
Field-based, biomechanically focused training interventions (Table 2.3) have shown the
combined training effects of plyometric, balance, resistance, and/or technique training,
implemented adjunct to an athlete’s normal in-season training, are effective in reducing
ACL injury risk during double-leg landing, double-leg stop-landing and sidestepping
tasks (Chappell & Limpisvasti, 2008; Lim et al., 2009; Zebis et al., 2008). Lim et al.
(2009; Y. S. Lee, personal communication, January 19, 2011) reported reductions in
peak extension and valgus knee moments as well as elevated hamstring-quadriceps
co-contraction during the WA phase of double-leg landing. Chappell and Limpisvasti
(2008) reported reductions in valgus knee moments during a double-leg stop-landing
task. Zebis et al. (2008) reported increases in medial hamstring muscle activation in the
pre-contact phase of sidestepping. However, because external knee loading was not
measured (Zebis et al., 2008), it is unclear if the observed changes in hamstring
activation are in response to training or increases/changes in knee joint loading. It
should also be noted that all of these studies were conducted in “ideal” training
environments with high athlete compliance, which was as high as 100% (Chappell &
Limpisvasti, 2008) and low coach-to-athlete ratios during training, which were
approximately 3:11 (Lim et al., 2009) and 2:33 (Chappell & Limpisvasti, 2008).
When tested in isolation, resistance training was not effective in reducing external knee
loading and subsequent ACL injury risk during pre-planned double-leg landing or
sidestepping (Cochrane et al., 2010; Herman et al., 2008). Conversely, both balance
and technique training, tested in isolation, have been proven effective in reducing peak
valgus (Cochrane et al., 2010; Dempsey et al., 2009) and internal rotation (Cochrane et
al., 2010) knee moments during pre-planned and unplanned sidestepping. Although
providing more clarity as to which training interventions most influence external knee
loading, again, the aforementioned training interventions were all performed under
“ideal” training settings (Cochrane et al., 2010; Dempsey et al., 2009; Hewett et al.,
1996; Myer et al., 2006; Myer et al., 2005).
To our knowledge, the efficacy and effectiveness of plyometric, balance, resistance,
and/or technique training in reducing the biomechanical factors associated with ACL
injury risk have not been tested in “real-world” settings, using a randomized control trial
(RCT) design. Future research is needed to fill this gap if positive laboratory-based
findings can be translated to “real-world” community level training environments.
26
Additionally, it is important to continue to test training interventions in isolation, but also
to identify the causal links between specific training classifications and surrogate
biomechanical measures of ACL injury risk. From both approaches, we will be better
able to develop ACL injury prevention training protocols that target the factors
associated with ACL injury risk and increase the probability of transferring positive
laboratory-based training effects to “real-world” training environments.
27
Table 2.1
Author n Training Protocol Results Interpretation
Gilchrist et al. (2008)
1,435(F) 1. Plyometric, Technique and Resistance (PTR) (n = 583)
a. See Mandelbaum et al., 2005. 2. Control (C)(n = 852)
a. Normal Training
Exposure: 1-yr; Regular season: 12 wks, 3 days per week, 20 min per day.
1. No significant differences in total non-contact ACL injuries were reported between the PTR and C groups.
Report of low statistical power (via # of non-contact ACL injuries)
2. Incidence of non-contact ACL injuries occurring during training (p = 0.014) and the second half of a seson (p = 0.025) per 1,000 players was reduced in the PTR group relative to the C group.
1. PTR training was not effective in decreasing total non-contact ACL injuries in female athletes.
2. PTR training may be most effective in reducing incidence of non-contact ACL injuries occurring during training and the second half of a season.
Steffen et al. (2008)
2020(F) 1. Plyometric, Balance, Technique and Resistance (PBTR)(n = 1,073)
Plyometric a. Double and single-leg b. Jumping and bounding
Balance c. Single-leg, unstable surfaces and perturbation
Technique (Landing)
d. “Core stability” (Trunk control) e. Minimize knee valgus
Resistance f. Lower body and trunk (‘core”)
2. Control (C)(n = 947) a. Normal Training
Exposure: 1-yr; Pre-season: 8 wks, 15 consecutive days, then 20 min per day, 1 day per wk. Regular season: 21 wks, 20 min per day, 1 days per wk.
1. No significant differences in ACL injuries were reported between the PBTR and C groups.
1. PBTR training was not effective in decreasing non-contact ACL injuries in female athletes.
M – male, F – female
ACL injury focused training interventions
28
Table 2.1 continued: ACL injury focused training interventions
Author n Training Protocol Results Interpretation
Pfeiffer et al. (2006)
1,439(F) 1. Plyometric and Technique (PT)(n = 577) Plyometric
a. Double and single-leg jump training Technique (Running, Sidestepping and Landing)
b. Alignment of hip, knee and ankle 2. Control (C) (n = 862)
a. Normal Training
Exposure: 2-yrs; Regular season: 2 days per wk, 20 min per day (number of wks not reported).
1. No significant differences in non-contact ACL injuries were reported between the PT and C groups.
1. PT training was not effective in decreasing non-contact ACL injuries in female athletes.
Mandelbaum et al. (2005)
5,703(F) 1. Year 1: Plyometric, Technique and Resistance (PTR) (n = 1,041)
Plyometric
a. Jumping, bounding and landing Technique (Landing)
b. Increase hip and knee flexion Resistance
c. Lower body 2. Year 1: Control (C) (n = 1,905)
a. Normal Training 3. Year 2: Plyometric, resistance and technique (PRT) (n = 844)
a. See year 1 4. Year 2: Control (C)(n = 1,913)
a. Normal Training
Exposure: 2-yrs; Regular season: 20 min per day (number of wks and days per wk not reported).
1. Incidence of non-contact ACL injuries per 1,000 players was reduced in the PTR group relative to the C group following year 1 (p = 0.001) and 2 (p = 0.005).
1. PTR training was effective in reducing incidence of non-contact ACL injuries in female athletes.
M – male, F – female
29
Table 2.1 continued: ACL injury focused training interventions
Author n Training Protocol Results Interpretation
Myklebust et al. (2003)
942(F) 1. Year 1: Control (C)(n = 942) a. Normal training
2. Year 2: Balance (B)(n = 855) Balance
a. Double to single-leg support b. Progression , unstable surfaces to unstable
surfaces with perturbations
Coaches were consulted and training protocol amended to included ‘sport specific training’, supervision and feedback from physical therapists. 3. Year 3: Balance, Plyometric/Technique (BPT) (n = 850)
Balance a. See year 2
Plyometric/Technique (Standing, Sidestepping , Jumping and Landing)
b. “Dynamic knee control”, “knee over toe”, control of core (trunk).
Exposure: 2-yrs; Regular season: first 5-7 wks, 15 min per day, 3 times per wk then 1 time per wk for remaining 21-23 wks.
1. No significant difference in ACL injuries were observed between the first (C), second (B) or third (BPT) intervention seasons.
2. In year 3 (BPT) ACL injury rates for elite level athletes were reduced if they completed the injury prevention program compared to those who did not (p = 0.001).
1. B or BPT training was not effective in decreasing ACL injuries in female athletes.
2. Athlete compliance may be an important factor associated with ACL injury prevention programs in elite level athletes.
Wedderkopp et al. (2003).
163(F) 1. Balance and Resistance (BR)(n = 77) Balance
a. Unstable surfaces (details not reported) Resistance, functional strength
b. Upper body, lower body and trunk 2. Functional strength (FS)(n = 86)
a. Upper body, lower body and trunk
Exposure: 1-yr; Regular season: 36 wks, 15-20 min per day (days per wk not reported).
1. No significant difference in ACL injuries were reported between the BR and FS groups.
Report of low statistical power (incidences of ACL injuries within each group)
1. Inconclusive evidence to support the use of BR to reduce ACL injury rates.
M – male, F – female
30
Table 2.1 continued: ACL injury focused training interventions
Author n Training Protocol Results Interpretation
Junge et al. (2002)
194(M) 1. Cardiovascular, Balance and Resistance (CBR) (n = 101)
Cardiovascular a. Heidt et al., (2000) (Details not reported)
Balance b. Caraffa et al., (1996) (Details not reported)
Resistance c. Lower body and trunk
2. Control (C)(n = 93) a. Normal Training
Exposure: 1-yr; Regular season (number of wks, min per day and days per week not reported).
1. No significant difference in ACL injuries were reported between the CBR and C groups.
Report of low statistical power (incidences of ACL injuries within each group)
1. Inconclusive evidence to support the use of CBR to reduce ACL injury rates.
Sӧderman et al. (2000)
140(F) 1. Balance (n = 62) a. Single-leg balance b. Progression, unstable surfaces to unstable surfaces with perturbations.
2. Control (n = 78) a. Normal Training
Exposure: 1-yr; Pre-season: 30 straight days for 10-15 min per day; Regular season: 3 days per wk, 10-15 min per day (Number of wks not reported).
1. No significant difference in ACL injuries were observed between the B and C groups.
Report of low statistical power (incidences of ACL injuries within each group)
1. B training was not effective in decreasing ACL injuries in female athletes.
M – male, F – female
31
Table 2.1 continued: ACL injury focused training interventions
Author n Training Protocol Results Interpretation
Heidt et al. (2000)
300(F) 1. Cardiovascular, Plyometric, Technique and Resistance (CPTR)(n = 42)
Cardiovascular
a. Acceleration on treadmill, incline and flat surfaces
Plyometric b. Progression, unilateral to multidirectional on compliant surfaces (details not reported)
Technique (Running) c. “sport coordination” (details not reported)
Resistance d. Details not reported
2. Control (C)(n = 258) a. Normal Training
Exposure: 1-yr; Pre-season: 7 wks, 2.86 days per week (duration per day not reported).
1. No significant difference in ACL injuries were observed between the CPTP and C groups.
1. CPTR training was not effective in decreasing ACL injuries in female athletes.
Hewett et al. (1999)
829(F) 434(M)
1. Plyometric, Technique and Resistance (PTR)(n = 366) Plyometric:
a. Double and single-leg jumping Technique (Landing):
b. Straight trunk & minimise sway c. Chest over knees e. Increase lower limb flexion
Resistance: f. Upper body, lower body and trunk
2. Control Female (CF)(n = 463) a. Normal training
3. Control Male (CM)(n = 434) a. Normal training
Exposure: 1-yr; Pre-season: 6-wks, 60-90 min per day, 3 days per wk.
1. Incidence of non-contact ACL injuries per 1,000 players was reduced (p = 0.05) in the PTR group relative to the CF group.
2. No significant differences in non-contact ACL injuries were reported between CF and CM groups.
1. BTR training was effective in reducing incidence of non-contact ACL injuries in female athletes.
M – male, F – female
32
Table 2.1 continued: ACL injury focused training interventions
Author n Training Protocol Results Interpretation
Caraffa et al. (1996)
600(M) 1. Balance (B) (n = 300). Balance
a. Double and single-leg, stable and unstable surfaces
2. Control (C)(n = 300) a. Normal training
Exposure: 3-yrs; Pre-season: ≥ 30 days, 20 min per day (days per wk not reported); Regular season: 20 min per day, 3 days per wk (number of wks not reported).
1. Incidence of ACL injury per team/season was reduced (p < 0.001) in the BT group relative to the C group.
Note: incidence of contact and non-contact ACL injuries not reported
1. B training was effective in reducing incidence of ACL injuries in male athletes.
M – male, F – female
33
Table 2.2
Author n Sport Task Training Protocol Results Interpretation
Cochrane et al. (2010)
50(M) 1. Pp SS 2. Un SS
1. Balance (B)(n = 10) a. Double and single-leg, unstable surfaces
2. Resistance; Machine Weights (RMW)(n = 10) a. Knee flexion and extension
3. Resistance; Free Weights (RFW)(n = 10) a. Knee flexion and extension
4. RMW + RBT(n = 10) 5. Control (C)(n = 10)
a. No training
EXPOSURE:12-wks, 30 min per day, 3 times per wk.
1. B: 62% ↓ peak valgus (N∙m-
1∙kg
-1), 24% ↓ varus (N∙m
-
1∙kg
-1), 32% ↓ int. rot. (N∙m
-
1∙kg
-1)
2. RMW: 27 % ↓ peak valgus (N∙m
-1∙kg
-1), 21% ↓ peak
varus (N∙m-1
∙kg-1
) 3. RFW: ↑ peak valgus, varus
and int. rot. knee moments (non-significant).
4. MW + B: minimal change. 5. C: 27 % ↑ peak valgus (N∙m
-
1∙kg
-1)
1. B training in isolation produced the most beneficial changes to knee joint loading and subsequent ACL injury risk during both Pp SS and Un SS in male athletes.
2. RFW training in isolation produced the least beneficial changes to knee loading and subsequent ACL injury risk during both Pp and Un SS in male athletes
Dempsey et al. (2009)
9(M) 1. Pp SS 2. Un SS
1. Technique (T) (Sidestepping) a. Bring stance foot towards midline b. Prevent “toeing” of stance foot c. Upright torso d. Rotate torso towards direction of travel
EXPOSURE:6-wks, 15 min per day, 2 times per wk.
1. T: 37% and 35% ↓ peak valgus (N∙m
-1∙kg
-1) during Pp
SS and Un SS respectively.
1. T training in isolation produced beneficial changes to knee joint loading and subsequent ACL injury risk during both Pp and Un SS in male athletes.
Herman et al. (2008)
66(F) 1. Pp DLL 1. Resistance (R) (n = 33) a. Knee flexion and extension b. Hip abduction and extension
2. Control (C) (n = 33) a. No resistance training
EXPOSURE:9-wks, 45 min per day, 3 times per wk.
1. No change in knee valgus and flexion moments, anterior tibial shear force or knee flexion angle.
1. R training in isolation was not effective in changing knee joint loading, kinematics and subsequent ACL injury risk during Pp DLL in female athletes.
Unless otherwise stated, all variables reported are during either during the weight acceptance phase of stance (first 30%) or the stance phase of landing immediately before maximum vertical jump. M – male, F – female, Pp – pre-planned, Un – Unplanned, SS – Sidestep, DLL – double-leg landing, SLL – single-leg landing.
Laboratory-based, biomechanically-focused training interventions
34
Table 2.2 continued: Laboratory-based, biomechanically-focused training interventions
Author n Sport Task Training Protocol Results Interpretation
Myer et al. (2006)
18(F) 1. Pp DLL 2. Pp SLL
1. Plyometric, Technique and Resistance (PTR)(n = 8) Plyometric:
a. Single and double-leg jump training Technique (Landing):
b. Minimize knee valgus motion c. Myer et al., (2005).
Resistance: d. Upper body, lower body and trunk
2. Balance, Technique and Resistance (BTR)(n = 10) Balance:
a. Myer et al., (2005). Technique (Landing):
b. Minimize knee valgus motion c. (Myer et al., 2005)
Resistance:
d. Upper body, lower body and trunk
EXPOSURE:6-wks, 90 min per day, 3 times per wk.
1. PTR: ↓ peak knee valgus (deg) during Pp SLL
2. PTR: ↑ peak knee flexion (deg) during Pp DLL
3. BTR: ↓ peak knee valgus (deg) during Pp SLL
4. BTR: ↑ peak knee flexion (deg) during Pp SLL
1. The combined effects of PTR produced beneficial knee joint kinematic changes and may reduce ACL injury risk during both Pp DLL and SLL tasks in female athletes.
Myer et al. (2005)
53(F) 1. Pp DLL 1. Plyometric, Balance, Technique, and Resistance (PBTR)(n = 41)
Plyometric: a. Single and double-leg jump training
Technique (Landing): b. “Dynamic control” - ↓ trunk sway c. Chest over knees - ↑ lower limb flexion
Balance: d. Double and single-leg, unstable surfaces
Resistance: e. Upper body, lower body and trunk
2. Control (C)(n = 12) a. No training
EXPOSURE:6-wks, 90 min per day, 3 times per wk.
1. PBTR: 28% and 38% ↓ in right knee peak varus/valgus respectively (N∙m
-1∙kg
-
1). Similar trends
observed for the left knee.
2. PBTR: ↑ right and left knee flexion ROM (deg)
3. C: No change in knee valgus or knee flexion ROM (deg).
1. The combined effects of PBTR training produced beneficial changes to knee joint loading and subsequent ACL injury risk during Pp DLL in female athletes.
2. The combined effects of PBTR training produced beneficial knee joint kinematic changes may reduce ACL injury risk during Pp DLL landing in female athletes.
Unless otherwise stated, all variables reported are during either during the weight acceptance phase of stance (first 30%) or the stance phase of landing immediately before maximum vertical jump. M – male, F – female, Pp – pre-planned, Un – Unplanned, SS – Sidestep, DLL – double-leg landing, SLL – single-leg landing.
35
Table 2.2 continued: Laboratory-based, biomechanically-focused training interventions
Author n Sport Task Training Protocol Results Interpretation
Hewett et al. (1996)
11(F)
1. Pp DLL 1. Plyometric, Technique and Resistance (PTR) Plyometric:
a. Double and single-leg jump training Technique (Landing):
b. Straight trunk c. Chest over knees d. Minimize body sway e. Increase lower limb flexion
Resistance: f. Upper body, lower body and trunk
EXPOSURE:6-wks, 120 min per day, 3 times per wk.
1. PTR: Varus dominant(n = 7), 38% ↓ peak varus (N∙m
-1∙kg
-1)
2. PTR: Valgus dominant (n = 4), 53% ↓ valgus moments (N∙m
-1∙kg
-1).
1. The combined effects of PTR training reduced ACL injury risk by reducing either peak varus or varus moments during Pp DLL in female athletes.
Unless otherwise stated, all variables reported are during either during the weight acceptance phase of stance (first 30%) or the stance phase of landing immediately before maximum vertical jump. M – male, F – female, Pp – pre-planned, Un – Unplanned, SS – Sidestep, DLL – double-leg landing, SLL – single-leg landing.
36
Table 2.3
Author n Sport Task Training Protocol Results Interpretation
Lim et al. (2009)
22(F) 1. Pp DLL
1. Plyometric, Technique and Resistance (PTR) (n = 11)
Plyometric: a. Single and double-leg jump training.
Technique (Landing): Details not reported
Resistance: b. Trunk and lower body
2. Control (C)(n = 11) a. Normal in-season training.
EXPOSURE: 8-wks, 20 min per day (days per wk not reported).
1. PTR: ↑ peak knee flexion (deg)
2. PTR: ↑ Ham/Quad co-contraction over time
3. C: No differences over time 4. Post training, peak valgus
(Nm) was 76% lower in the PTR group when compared to the control group.
1. PTR training adjunct to normal in-season training reduced ACL injury risk by reducing peak valgus moments and increasing Ham/Quad co-contraction during Pp DLL in female athletes.
2. PTR training adjunct to normal in-season training produced beneficial knee joint kinematic changes and may reduce ACL injury risk during Pp DLL in female athletes.
Chappell & Limpisvasti (2008)
30(F) 1. Pp DLL 2. Pp
double-leg stop-landing
1. Plyometric/Balance and Resistance (PBR) Plyometric/Balance:
a. Double and single-leg jump training b. Single-leg, stable surfaces, perturbation
Resistance: c. Trunk and lower body
EXPOSURE:6-wks, 15-20 min per day, 6 days per wk.
1. PBR: 15% ↓ peak valgus (Nm) during Pp double-leg stop-landing
2. PBR: 21% ↓ peak knee flexion (Nm) during Pp DLL
3. PBR: ↑ peak knee flexion (deg) during Pp DLL
1. PBR training adjunct to in-season training reduced peak valgus knee moments and subsequent ACL injury risk during Pp double-leg stop landing in female athletes.
2. PBR training adjunct to in-season training reduced flexion knee moments and elevated knee flexion angle, which may reduce ACL injury risk during Pp DLL in female athletes.
Unless otherwise stated, all variables reported are during either during the weight acceptance phase of stance (first 30%) or the stance phase of landing immediately before maximum vertical jump. M – male, F – female, Pp – pre-planned, Un – Unplanned, SS – Sidestep, DLL – double-leg landing, SLL – single-leg landing.
Field-based, biomechanically-focused training interventions
37
Table 2.3 continued: Field-based, biomechanically-focused training interventions
Author n Sport Task Training Protocol Results Interpretation
Zebis et al. (2008)
20(F) 1. Pp SS 1. Balance and Technique (BT)(n = 20) Balance:
a. Unstable surfaces Technique (Standing, Running, Sidestepping, Jumping and Landing):
b. “dynamic control” of hip, knee and ankle 2. Control (C)(n = 20)
a. Normal training
EXPOSURE: BT, 52-wks, 20 min per day, 2 days per week. C, 24-wks (min per day and days per wk not reported).
1. BT: ↑ preparatory medial hamstring muscle activity (% Max).
2. BT: ↓ preparatory gluteus medius muscle activity (% Max).
3. BT: ↑ medial hamstring muscle activity (% Max) within WA.
4. BT: ↓ biceps femoris and gluteus medius muscle activity (% Max) within WA.
1. BT training adjunct to in-season training can elevated preparatory medial hamstring muscle activity, supporting the knee from external valgus and flexion knee loading, reducing ACL injury risk during Pp SS in female athletes.
Chimera et al. (2004)
18(F) 1. Pp DLL 1. Plyometric and Resistance (PR)(n = 9) Plyometric:
a. Double-leg jump training Resistance:
Details not reported 2. Resistance (R)(n = 9)
Details not reported
EXPOSURE: 6-wks, 2 days per week (min per day not reported). Addition of 20-30 min of plyometric training for PR group.
1. PR: ↑ preparatory hip adductor muscle activity (% Max).
2. PR: ↑ preparatory hip ad/abductor co-contraction
1. PR training adjunct to normal pre-season training can elevated preparatory hip adductor muscle activation, stabilizing the hip and may indirectly support the knee from external loading and reduce ACL injury risk during Pp DLL in female athletes.
Unless otherwise stated, all variables reported are during either during the weight acceptance phase of stance (first 30%) or the stance phase of landing immediately before maximum vertical jump. M – male, F – female, Pp – pre-planned, Un – Unplanned, SS – Sidestep, DLL – double-leg landing, SLL – single-leg landing.
38
2.9 ATHLETE SCREENING
Injury prevention training programs have been shown to have greater effects on “high-risk”
relative to “low-risk” athletic populations (Myer, Ford, Brent, & Hewett, 2007). The ability to
use screening tools to identify “high-risk” athletic populations would provide health care
professionals with the ability to develop athlete-specific ACL injury prevention training
protocols. Tibial-femoral bone geometry, such as narrow intercondylar notch width and a
steep posterior tibial slope angle, have both been associated with elevated ACL injury risk
(Simon, Everhart, Nagaraja, & Chaudhari, 2010). The ability to image an athlete’s tibial
morphology is expensive and is a non-modifiable risk factor, limiting the use of this type of
screening tool in community level athletic populations.
Clinically relevant tests for estimating ACL injury risk have shown that medial/lateral upper
body stability is the strongest single predictor of ACL injury (Overall p = 0.02; Odds Ratio =
2.2) (Zazulak, Hewett, Reeves, Goldberg, & Cholewicki, 2007), while whole body postural
stability is one of four predictors of ACL reinjury (C statistic = 0.94; Paterno et al., 2010).
Although more accessible, cost effective, and modifiable, these clinically relevant
screening tools provide limited causal information between their predictive measures and
the biomechanical factors associated with ACL injury risk. If available, this information can
be used to prescribe personalized injury prevention training protocols that target the
biomechanics factors classifying an athlete as “high-risk” as well as maximizing the impact
of a prophylactic training protocol.
2.10 SUMMARY
Figure 2.4 presents an injury prevention framework specific to and detailed for non-contact
ACL injuries. In summary, the injury surveillance literature has shown that the majority of
sport-related ACL injuries occur during non-contact sidestepping and single-leg landing
tasks (Stage 1). Combined externally applied flexion, valgus, and internal rotation knee
moments during the WA phase of sidestepping and single-leg landing with the knee near
full extension is the likely mechanism of noncontact ACL injuries (Stage 2).
Countermeasures to reduce the biomechanical factors associated with ACL injury risk
should have three foci: (1) to reduce the magnitude of externally applied flexion, valgus,
and internal rotation knee moments; (2) to increase muscular support against these
aforementioned joint moments; and (3) to increase knee flexion angle and the
39
neuromuscular control of the hip during the WA phase of sidestepping and single-leg
landing; although the extent of (3) is still to be defined (Stage 3).
Figure 2.4
The combined effects of plyometric, balance, resistance, and/or technique training have
been effective in reducing the biomechanical factors associated with ACL injury in “ideal”
training environments (Stage 4). Literature testing the efficacy of ACL injury prevention
protocols largely remains in stage 4. We now need to test the efficacy of these training
interventions in “real-world” settings using an RCT design to determine if positive
laboratory-based biomechanical training outcomes, like reducing peak knee loading and/or
increasing muscular support, can be effectively translated to community level training
environments (Stage 5). Future research is also needed to evaluate the challenges
associate with implementing effective “real-world” training interventions within community
level training environments (Stage 6). The overall goal of such evaluations will be to
observe reductions in ACL injury rates across heterogeneous community level athletic
Aetiology
1. In-lab
2. In-vivo/ Cadaveric
3. In-silico
(Stage 2)
Countermeasure Development
(Stage 3)
Injury Surveillance
(Stage 1)
Training Intervention
‘Ideal’ Scenario
(Stage 4)
Training Intervention
‘Real-World’ Scenario (RCT)
(Stage 5)
Athlete
Screening
Community Level Adoption &
Maintenance
(Stage 6)
ACL injury prevention framework to translate ACL focused research into injury prevention practice.
40
populations (Stage 1). It should be noted that continuous, reliable, nationwide annual
injury surveillance systems are needed before the long-term effectiveness and cost
benefits of “real-world” ACL injury prevention training protocols in reducing ACL injury
rates can be evaluated.
Adding to the TRIPP model (Finch, 2006), the ACL injury prevention framework includes
athlete screening to identify “high-risk” athletes, allowing for the development of athlete-
specific ACL injury prevention training protocols. Finally, it is evident that the use of
feedback within the ACL injury prevention framework is needed to determine how
biomechanically relevant risk factors like peak joint loading and muscular support are
influenced following a training intervention and/or during athlete screening. By identifying
these causal relationships, ACL injury prevention training programs can be created to
target the biomechanically relevant factors associated with ACL injury risk in both general
and “high-risk” athletic populations. It is through this approach that more effective injury
prevention training programs can be developed and, in turn, ACL injury rates can be
reduced in the future.
Acknowledgements
The authors wish to thank the Australian National Health and Medical Research Council
(grant numbers 400937, 565907), the Australian football Research and Development
Board, the Western Australian Medical Health and Research Infrastructure Fund, The
Canadian Society for Biomechanics and The University of Western Australia for financial
support. CFF is supported by an NHMRC Principal Research Fellowship (565900).
Reference list chapter 2
Amiel, D., Kuiper, S. D., Wallace, C. D., Harwood, F. L., & VandeBerg, J. S. (1991). Age-
related properties of medial collateral ligament and anterior cruciate ligament: a morphologic and collagen maturation study in the rabbit. J Gerontol, 46(4), B159-165.
Besier, T. F., Lloyd, D. G., Ackland, T. R., & Cochrane, J. L. (2001). Anticipatory effects on
knee joint loading during running and cutting maneuvers. Med Sci Sports Exerc, 33(7), 1176-1181.
Besier, T. F., Lloyd, D. G., Cochrane, J. L., & Ackland, T. R. (2001). External loading of the
knee joint during running and cutting maneuvers. Med Sci Sports Exerc, 33(7), 1168-1175.
41
Buchanan, T. S., & Lloyd, D. G. (1997). Muscle activation at the human knee during isometric flexion-extension and varus-valgus loads. J Orthop Res, 15(1), 11-17.
Buford, W. L., Jr., Ivey, F. M., Jr., Nakamura, T., Patterson, R. M., & Nguyen, D. K. (2001).
Internal/external rotation moment arms of muscles at the knee: moment arms for the normal knee and the ACL-deficient knee. Knee, 8(4), 293-303.
Caraffa, A., Cerulli, G., Projetti, M., Aisa, G., & Rizzo, A. (1996). Prevention of anterior
cruciate ligament injuries in soccer. A prospective controlled study of proprioceptive training. Knee Surg Sports Traumatol Arthrosc, 4(1), 19-21.
Cerulli, G., Benoit, D. L., Lamontagne, M., Caraffa, A., & Liti, A. (2003). In vivo anterior
cruciate ligament strain behaviour during a rapid deceleration movement: case report. Knee Surg Sports Traumatol Arthrosc, 11(5), 307-311.
Chappell, J. D., & Limpisvasti, O. (2008). Effect of a neuromuscular training program on
the kinetics and kinematics of jumping tasks. Am J Sports Med, 36(6), 1081-1086. Chaudhari, A. M., Hearn, B. K., & Andriacchi, T. P. (2005). Sport-dependent variations in
arm position during single-limb landing influence knee loading: implications for anterior cruciate ligament injury. Am J Sports Med, 33(6), 824-830.
Cochrane, J. L., Lloyd, D. G., Besier, T. F., Elliott, B. C., Doyle, T. L., & Ackland, T. R.
(2010). Training affects knee kinematics and kinetics in cutting maneuvers in sport. Med Sci Sports Exerc, 42(8), 1535-1544.
Cochrane, J. L., Lloyd, D. G., Buttfield, A., Seward, H., & McGivern, J. (2007).
Characteristics of anterior cruciate ligament injuries in Australian football. J Sci Med Sport, 10(2), 96-104.
Dempsey, A. R., Lloyd, D. G., Elliott, B. C., Steele, J. R., & Munro, B. J. (2009). Changing
sidestep cutting technique reduces knee valgus loading. Am J Sports Med, 37(11), 2194-2200.
Dempsey, A. R., Lloyd, D. G., Elliott, B. C., Steele, J. R., Munro, B. J., & Russo, K. A.
(2007). The effect of technique change on knee loads during sidestep cutting. Med Sci Sports Exerc, 39(10), 1765-1773.
Dunn, W. R., & Spindler, K. P. (2010). Predictors of activity level 2 years after anterior
cruciate ligament reconstruction (ACLR): a Multicentre Orthopaedic Outcomes Network (MOON) ACLR cohort study. Am J Sports Med, 38(10), 2040-2050.
Finch, C.F. (2006). A new framework for research leading to sports injury prevention. J Sci
Med Sport, 9(1-2), 3-9; discussion 10. Fukuda, Y., Woo, S. L., Loh, J. C., Tsuda, E., Tang, P., McMahon, P. J., & Debski, R. E.
(2003). A quantitative analysis of valgus torque on the ACL: a human cadaveric study. Journal of orthopaedic research, 21(6), 1107-1112.
Gabriel, M. T., Wong, E. K., Woo, S. L., Yagi, M., & Debski, R. E. (2004). Distribution of in
situ forces in the anterior cruciate ligament in response to rotatory loads. J Orthop Res, 22(1), 85-89.
42
Gianotti, S. M., Marshall, S. W., Hume, P. A., & Bunt, L. (2009). Incidence of anterior cruciate ligament injury and other knee ligament injuries: a national population-based study. J Sci Med Sport, 12(6), 622-627.
Granan, L. P., Forssblad, M., Lind, M., & Engebretsen, L. (2009). The Scandinavian ACL
registries 2004-2007: baseline epidemiology. Acta Orthop, 80(5), 563-567. Hashemi, J., Breighner, R., Jang, T. H., Chandrashekar, N., Ekwaro-Osire, S., &
Slauterbeck, J. R. (2010). Increasing pre-activation of the quadriceps muscle protects the anterior cruciate ligament during the landing phase of a jump: an in vitro simulation. Knee, 17(3), 235-241.
Heidt, R. S., Jr., Sweeterman, L. M., Carlonas, R. L., Traub, J. A., & Tekulve, F. X. (2000).
Avoidance of soccer injuries with preseason conditioning. Am J Sports Med, 28(5), 659-662.
Herman, D. C., Weinhold, P. S., Guskiewicz, K. M., Garrett, W. E., Yu, B., & Padua, D. A.
(2008). The effects of strength training on the lower extremity biomechanics of female recreational athletes during a stop-jump task. Am J Sports Med, 36(4), 733-740.
Hewett, T. E., Lindenfeld, T. N., Riccobene, J. V., & Noyes, F. R. (1999). The effect of
neuromuscular training on the incidence of knee injury in female athletes. A prospective study. Am J Sports Med, 27(6), 699-706.
Hewett TE, Myer GD, Ford KR, Heidt RS, Jr., Colosimo AJ, McLean SG, van den Bogert
AJ, Paterno MV, and Succop P. (2005). Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes: a prospective study. Am J Sports Med. 33(4), 492-501.
Hewett, T. E., Stroupe, A. L., Nance, T. A., & Noyes, F. R. (1996). Plyometric training in
female athletes. Decreased impact forces and increased hamstring torques. Am J Sports Med, 24(6), 765-773.
Janssen, K. W., Orchard, J. W., Driscoll, T. R., & van Mechelen, W. (2011). High incidence
and costs for anterior cruciate ligament reconstructions performed in Australia from 2003-2004 to 2007-2008: time for an anterior cruciate ligament register by Scandinavian model? Scand J Med Sci Sports. doi: 10.1111/j.1600-0838.2010.01253.x
Jindrich, D. L., Besier, T. F., & Lloyd, D. G. (2006). A hypothesis for the function of braking
forces during running turns. J Biomech, 39(9), 1611-1620. Junge, A., Rosch, D., Peterson, L., Graf-Baumann, T., & Dvorak, J. (2002). Prevention of
soccer injuries: a prospective intervention study in youth amateur players. Am J Sports Med, 30(5), 652-659.
Kipp, K., McLean, S. G., & Palmieri-Smith, R. M. (2011). Patterns of hip flexion motion
predict frontal and transverse plane knee torques during a single-leg land-and-cut maneuver. Clin Biomech (Bristol, Avon), 26(5), 504-508.
43
Koga, H., Nakamae, A., Shima, Y., Iwasa, J., Myklebust, G., Engebretsen, L., . . . Krosshaug, T. (2010). Mechanisms for noncontact anterior cruciate ligament injuries: knee joint kinematics in 10 injury situations from female team handball and basketball. Am J Sports Med, 38(11), 2218-2225.
Krosshaug, T., Nakamae, A., Boden, B. P., Engebretsen, L., Smith, G., Slauterbeck, J. R.,
. . . Bahr, R. (2007). Mechanisms of anterior cruciate ligament injury in basketball: video analysis of 39 cases. Am J Sports Med, 35(3), 359-367.
Li, G., Rudy, T. W., Sakane, M., Kanamori, A., Ma, C. B., & Woo, S. L. (1999). The
importance of quadriceps and hamstring muscle loading on knee kinematics and in-situ forces in the ACL. J Biomech, 32(4), 395-400.
Lim, B. O., Lee, Y. S., Kim, J. G., An, K. O., Yoo, J., & Kwon, Y. H. (2009). Effects of
sports injury prevention training on the biomechanical risk factors of anterior cruciate ligament injury in high school female basketball players. Am J Sports Med, 37(9), 1728-1734.
Lloyd, D. G. (2001). Rationalee for training programs to reduce anterior cruciate ligament
injuries in Australian football. J Orthop Sports Phys Ther, 31(11), 645-654; discussion 661.
Lloyd, D. G., & Buchanan, T. S. (1996). A model of load sharing between muscles and soft
tissues at the human knee during static tasks. J Biomech Eng, 118(3), 367-376. Lloyd, D. G., & Buchanan, T. S. (2001). Strategies of muscular support of varus and
valgus isometric loads at the human knee. J Biomech, 34(10), 1257-1267. Lloyd, D. G., Buchanan, T. S., & Besier, T. F. (2005). Neuromuscular biomechanical
modelling to understand knee ligament loading. Med Sci Sports Exerc, 37(11), 1939-1947.
Lyman, S., Koulouvaris, P., Sherman, S., Do, H., Mandl, L. A., & Marx, R. G. (2009).
Epidemiology of anterior cruciate ligament reconstruction: trends, readmissions, and subsequent knee surgery. J Bone Joint Surg Am, 91(10), 2321-2328.
Mandelbaum, B. R., Silvers, H. J., Watanabe, D. S., Knarr, J. F., Thomas, S. D., Griffin, L.
Y., . . . Garrett, W., Jr. (2005). Effectiveness of a neuromuscular and proprioceptive training program in preventing anterior cruciate ligament injuries in female athletes: 2-year follow-up. Am J Sports Med, 33(7), 1003-1010.
Markolf, K. L., Burchfield, D. M., Shapiro, M. M., Shepard, M. F., Finerman, G. A., &
Slauterbeck, J. L. (1995). Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res, 13(6), 930-935.
McGinley, J. L., Baker, R., Wolfe, R., & Morris, M. E. (2009). The reliability of three-
dimensional kinematic gait measurements: a systematic review. Gait Posture, 29(3), 360-369.
McLean, S. G., Borotikar, B., & Lucey, S. M. (2010). Lower limb muscle pre-motor time
measures during a choice reaction task associate with knee abduction loads during dynamic single leg landings. Clin Biomech (Bristol, Avon), 25(6), 563-569.
44
McLean, S. G., Huang, X., Su, A., & Van Den Bogert, A. J. (2004). Sagittal plane biomechanics cannot injure the ACL during sidestep cutting. Clin Biomech (Bristol, Avon), 19(8), 828-838.
McLean, S. G., Huang, X., & van den Bogert, A. J. (2005). Association between lower
extremity posture at contact and peak knee valgus moment during sidestepping: implications for ACL injury. Clin Biomech (Bristol, Avon), 20(8), 863-870.
McLean, S. G., Huang, X., & van den Bogert, A. J. (2008). Investigating isolated
neuromuscular control contributions to non-contact anterior cruciate ligament injury risk via computer simulation methods. Clin Biomech (Bristol, Avon), 23(7), 926-936.
McLean, S. G., & Samorezov, J. E. (2009). Fatigue-induced ACL injury risk stems from a
degradation in central control. Med Sci Sports Exerc, 41(8), 1661-1672. Meyer, E. G., & Haut, R. C. (2008). Anterior cruciate ligament injury induced by internal
tibial torsion or tibiofemoral compression. Journal of biomechanics, 41(16), 3377-3383.
More, R. C., Karras, B. T., Neiman, R., Fritschy, D., Woo, S. L., & Daniel, D. M. (1993).
Hamstrings--an anterior cruciate ligament protagonist. An in vitro study. Am J Sports Med, 21(2), 231-237.
Myer, G. D., Ford, K. R., Brent, J. L., & Hewett, T. E. (2007). Differential neuromuscular
training effects on ACL injury risk factors in "high-risk" versus "low-risk" athletes. BMC Musculoskelet Disord, 8(39), 1-7. doi: 1471-2474-8-39.
Myer, G. D., Ford, K. R., McLean, S. G., & Hewett, T. E. (2006). The effects of plyometric
versus dynamic stabilization and balance training on lower extremity biomechanics. Am J Sports Med, 34(3), 445-455.
Myer, G. D., Ford, K. R., Palumbo, J. P., & Hewett, T. E. (2005). Neuromuscular training
improves performance and lower-extremity biomechanics in female athletes. J Strength Cond Res, 19(1), 51-60.
Myklebust, G., Engebretsen, L., Braekken, I. H., Skjolberg, A., Olsen, O. E., & Bahr, R.
(2003). Prevention of anterior cruciate ligament injuries in female team handball players: a prospective intervention study over three seasons. Clin J Sport Med, 13(2), 71-78.
Noyes, F. R. (1977). Functional properties of knee ligaments and alterations induced by
immobilization: a correlative biomechanical and histological study in primates. Clin Orthop Relat Res(123), 210-242.
Noyes, F. R., & Grood, E. S. (1976). The strength of the anterior cruciate ligament in
humans and Rhesus monkeys. J Bone Joint Surg Am, 58(8), 1074-1082. Oiestad, B.E., Engebretsen, L., Storheim, K., Risberg, M.A., 2009. Knee osteoarthritis
after anterior cruciate ligament injury: A systematic review. Am J Sports Med, 37(7), 1434-1443.
45
Paterno, M. V., Schmitt, L. C., Ford, K. R., Rauh, M. J., Myer, G. D., Huang, B., & Hewett, T. E. (2010). Biomechanical measures during landing and postural stability predict second anterior cruciate ligament injury after anterior cruciate ligament reconstruction and return to sport. Am J Sports Med, 38(10), 1968-1978.
Pfeiffer, R. P., Shea, K. G., Roberts, D., Grandstrand, S., & Bond, L. (2006). Lack of effect
of a knee ligament injury prevention program on the incidence of noncontact anterior cruciate ligament injury. J Bone Joint Surg Am, 88(8), 1769-1774.
Rochcongar, P., Laboute, E., Jan, J., & Carling, C. (2009). Ruptures of the anterior
cruciate ligament in soccer. Int J Sports Med, 30(5), 372-378. Roos, H., Ornell, M., Gardsell, P., Lohmander, L. S., & Lindstrand, A. (1995). Soccer after
anterior cruciate ligament injury--an incompatible combination? A national survey of incidence and risk factors and a 7-year follow-up of 310 players. Acta Orthop Scand, 66(2), 107-112.
Shin, C. S., Chaudhari, A. M., & Andriacchi, T. P. (2011). Valgus plus internal rotation
moments increase anterior cruciate ligament strain more than either alone. Med Sci Sports Exerc, 43(8), 1484-1491.
Simon, R. A., Everhart, J. S., Nagaraja, H. N., & Chaudhari, A. M. (2010). A case-control
study of anterior cruciate ligament volume, tibial plateau slopes and intercondylar notch dimensions in ACL-injured knees. J Biomech, 43(9), 1702-1707.
Soderman, K., Werner, S., Pietila, T., Engstrom, B., & Alfredson, H. (2000). Balance board
training: prevention of traumatic injuries of the lower extremities in female soccer players? A prospective randomized intervention study. Knee Surg Sports Traumatol Arthrosc, 8(6), 356-363.
Steffen, K., Myklebust, G., Olsen, O. E., Holme, I., & Bahr, R. (2008). Preventing injuries in
female youth football--a cluster-randomized controlled trial. Scand J Med Sci Sports, 18(5), 605-614.
TheWorldBank. (June, 2010). World Population Estimates. Available from The World Bank
Group http://data.worldbank.org. Wedderkopp, N., Kaltoft, M., Holm, R., & Froberg, K. (2003). Comparison of two
intervention programmes in young female players in European handball--with and without ankle disc. Scand J Med Sci Sports, 13(6), 371-375.
Winter, D. (2005). Motor Control of Human Movement (3 ed.). Hoboken, New Jersey: John
Wiley & Sons, Inc. Withrow, T. J., Huston, L. J., Wojtys, E. M., & Ashton-Miller, J. A. (2008). Effect of varying
hamstring tension on anterior cruciate ligament strain during in vitro impulsive knee flexion and compression loading. J Bone Joint Surg Am, 90(4), 815-823.
Woo, S. L., Gomez, M. A., Sites, T. J., Newton, P. O., Orlando, C. A., & Akeson, W. H.
(1987). The biomechanical and morphological changes in the medial collateral ligament of the rabbit after immobilization and remobilization. J Bone Joint Surg Am, 69(8), 1200-1211.
46
Wu, J. L., Hosseini, A., Kozanek, M., Gadikota, H. R., Gill, T. J. t., & Li, G. (2010). Kinematics of the anterior cruciate ligament during gait. Am J Sports Med, 38(7), 1475-1482.
Wu, J. L., Seon, J. K., Gadikota, H. R., Hosseini, A., Sutton, K. M., Gill, T. J., & Li, G.
(2010). In situ forces in the anteromedial and posterolateral bundles of the anterior cruciate ligament under simulated functional loading conditions. Am J Sports Med, 38(3), 558-563.
Zazulak, B. T., Hewett, T. E., Reeves, N. P., Goldberg, B., & Cholewicki, J. (2007). Deficits
in neuromuscular control of the trunk predict knee injury risk: a prospective biomechanical-epidemiologic study. Am J Sports Med, 35(7), 1123-1130.
Zebis, M. K., Bencke, J., Andersen, L. L., Dossing, S., Alkjaer, T., Magnusson, S. P., . . .
Aagaard, P. (2008). The effects of neuromuscular training on knee joint motor control during sidecutting in female elite soccer and handball players. Clin J Sport Med, 18(4), 329-337.
47
CHAPTER 3
CHANGES IN KNEE JOINT BIOMECHANICS FOLLOWING BALANCE AND
TECHNIQUE TRAINING AND A SEASON OF AUSTRALIAN FOOTBALL
This manuscript has been accepted for publication in the British Journal of Sports
Medicine.
Donnelly C.J., Elliott, B.C., Doyle, T.L.A., Finch, C.F., Dempsey, A.R. and Lloyd, D.G. (2012). Changes in knee joint biomechanics following balance and technique training and a season of Australian football. Br J Sports Med. doi: 10.1136/bjsports-2011-090829. . The PhD candidate, Cyril J. Donnelly accounted for 70% of the intellectual property
associated with the final manuscript. Collectively, the remaining authors contributed 30%.
Conflict of Interest: There were no financial or personal relationships with other people or
organizations that could have biased the presented work
48
Abstract
Purpose Determine if balance and technique training (BTT) implemented adjunct to
normal Australian football (AF) training reduces external knee loading during sidestepping.
Additionally, the authors determined if an athlete’s knee joint kinematics and kinetics
change over a season of AF. Methodology Eight amateur-level AF clubs (n=1,001 males)
volunteered to participate in either 28 weeks of BTT or a ‘sham’ training (ST) adjunct to
their normal preseason and regular training. A subset of 34 athletes (BTT, n=20; ST,
n=14) were recruited for biomechanical testing in weeks 1–7 and 18–25 of the 28-week
training intervention. During biomechanical testing, participants completed a series
running, preplanned (PpSS) and unplanned sidestepping (UnSS) tasks. A linear mixed
model (α=0.05) was used to determine if knee kinematics and peak moments during PpSS
and UnSS were influenced by BTT and/or a season of AF. Results Both training groups
significantly (p=0.025) decreased their peak internal-rotation knee moments during PpSS,
and significantly (p=0.022) increased their peak valgus knee moments during UnSS
following their respective training interventions. Conclusions BTT was not effective in
changing an athlete’s knee joint biomechanics during sidestepping when conducted in
‘real-world’ training environments. Following normal AF training, the players had different
changes to their knee joint biomechanics during both preplanned and unplanned
sidestepping. When performing an unplanned sidestepping task in the latter half of a
playing season, athletes are at an increased risk of ACL injury. The authors therefore
recommend both sidestepping tasks are performed during biomechanical testing when
assessing the effectiveness of prophylactic training protocols.
KEYWORDS: KNEE; ACL; BIOMECHANICS; INJURY PREVENTION; TRAINING
3.1 INTRODUCTION
Anterior cruciate ligament (ACL) injuries in sport are common[1] and associated with high
financial and personal cost. In New Zealand and Australia, ACL injuries cost their
respective healthcare systems approximately 17.4 million NZD[1] and 75 million AUD[2]
per year. Following an ACL injury, over 50% of athletes are not capable of returning to the
same level of competition 2 years post reconstruction,[3] a percentage that increases to
approximately 70% after 3 years.[4]
49
One-half of non-contact ACL injuries occur during sidestepping sport tasks.[5]
Biomechanical analysis of sidestepping shows that compared with straight-line running,
peak extension knee moments are similar, while internal rotation and/or valgus knee
moments are elevated;[6–8] the same loading patterns that elevate ACL strain measured
in cadaveric knee models.[9] Peak in vivo ACL strain during sporting tasks characterised
by a rapid deacceleration phase,[10] like sidestepping,[11] generally occur during the
weight-acceptance (WA) phase of stance (first 20–30%)[7 8 12] and thought to be when
ACL injury risk is the greatest.
The ACL consists of two bundles, the anteromedial bundle (AMB) and posterolateral
bundle (PLB), named from their insertions on the tibial plateau. Modelling of the AMB and
PLB shows the kinematics of the ACL change as a function of knee flexion angle, with
peak elongation observed near full extension.[13] These results show that knee flexion
during stance is associated with ACL injury risk.
Reducing externally applied forces to the ACL can be achieved in two ways. First, reduce
the size of the loads applied to the knee by changing an individual’s posture or technique
during sidestepping.[12 14–16] Second, increase the strength and/or activation of the
muscles crossing the knee capable of protecting it when loads are elevated.[14 17]
Training interventions like balance[8] and technique training[15] have been shown to be
effective in reducing internal rotation and/or valgus knee moments during sidestepping.
However, these training interventions have only been shown to be effective when
implemented in ‘ideal’ settings,[8 15] which for this study is defined as a training
intervention conducted in a controlled laboratory setting, with high athlete compliance
(>80%) and a low coach to athlete ratio (<1:20). To date, no study has determined if
balance and technique is effective in reducing peak knee loading during sidestepping
when implemented in a ‘real-world’ training environment. This is where training is
conducted in a community-level setting with similar coach to athlete ratios as observed
during normal training and the instruction is given by a trainer blinded to the intended aims
and outcome measures of the intervention.
The primary purpose of this investigation was to determine if balance and technique
training (BTT), implemented adjunct to normal preseason and regular season AF training
reduces peak knee moments and/or influenced an athlete’s knee flexion angle during the
50
WA phase of pre-planned (PpSS) and unplanned (UnSS) sidestepping. Additionally, we
determined if an athlete’s knee joint biomechanics change over a season of AF. With this
information, we can establish if positive laboratory-based training outcomes can be
translated to ‘real-world’ community-level training environments, and if an athlete’s ACL
injury risk changes over a playing season.
3.2 METHODS This study was approved by the Human Research Ethics Committees at The University of
Western Australia (UWA) and the University of Ballarat.
3.2.1 Participant population – training intervention
As part of the Preventing Australian Football Injuries through Exercise (PAFIX) study,[19]
eight Western Australian Amateur Football League (WAAFL) clubs (n=1,001 males)
participated in either 28 weeks of BTT or a ‘sham’ training (ST) intervention adjunct to their
2007 or 2008 preseason and regular season training. The ST intervention served as the
experimental control group. All participants provided their informed, written consent before
participating in their respective training interventions.
3.2.2 Participant population – biomechanical testing
From an alphabetical list of the eligible WAAFL participants (n=1,001) 58 athletes were
recruited via a phone interview by an independent researcher 1 week before training
through the first 7 weeks of training (weeks 1–7) for biomechanical testing. Thirty-four
(59%) were available for follow-up testing in weeks 18 to 25 (BTT, n=14; ST, n=20) (figure
3.1). Exclusion criteria for participants included self-reported joint disorders or had
undergone an orthopaedic surgical procedure. All participants provided their informed,
written consent before biomechanical testing.
51
Figure 3.1
3.2.3 Training protocol
Two independent research assistants blinded to (1) which training programs they were
overseeing, and (2) the outcome variables analysed during biomechanical testing were
assigned to each of the eight WAAFL clubs. Each club consisted of three teams (grade A,
B and C), with approximately 25–30 players per team. This made a trainer to athlete ratio
of approximately 1:40 for each club. To run each training session, each club uses a staff
consisting of a head coach, assistant coaches and athletic trainers. Our research
assistants were used in place of the club’s normal athletic trainers to conduct 20 min of
either ST or BTT at the beginning of each club’s regularly scheduled training sessions.
Research assistants were considered qualified to run these training sessions after
completing a 20 h coaching seminar associated with their respected training intervention.
These research assistants also accurately recorded athlete participation following each
training session.[20]
Experimental data flow of training intervention and biomechanical testing sessions 1 and 2. BTT and ST numbers were only reported in testing session two as the biomechanists conducting the data collections were blinded to the training intervention codes of each participant until the statistics phase of the analysis. Mean ± standard deviation age, body mass and height were reported for participants who completed both testing session 1 and 2.
52
A WAAFL playing season consists of 8 weeks of preseason training and 20 weeks of
regular season training, with two regular season bye weeks, where teams trained but did
not play a match. Each club trained twice per week during the preseason and regular
season, and played one match a week during the regular season.
Training interventions (BTT or ST) were conducted for 20 min before each team’s normal
training, twice a week in the 8-week preseason, and during the first 10 weeks of the
regular season (18 weeks). Training was then condensed to once per week for weeks 19
through 28 (total training sessions, n=46). Of the participants that completed both
biomechanical testing sessions, athletes in the BTT group attended 45±22% of the total
training sessions and the ST group attended 51±33%. A one-way ANOVA showed no
differences (p=0.696) in the number of training sessions completed by athletes in the BTT
and ST groups.
The BTT protocol used for this study is an extension of previous training methods shown
to be effective in decreasing peak knee moments during sidestepping.8 15 Balance
training included single-leg, wobble board, stability disk and Swiss stability ball balance
tasks. Each balance exercise became progressively more difficult from weeks 1 to week
18 with the last 10 weeks of training designed as a maintenance phase. During each
training session, when appropriate, athletes were verbally instructed to keep their stance
foot close to midline, maintain a controlled vertical trunk posture and increase knee flexion
during the stance phase of both sidestepping and landing tasks. Interested readers can
obtain a detailed description of the BTT training protocol from the corresponding author
(Appendix A).
The primary goal of the ST intervention was for athletes to concentrate on improving their
acceleration during straight line running tasks, which to our knowledge has not been
shown to significantly decrease peak knee joint change to loading or ACL injury rates
following training. Neither technique instruction nor balance tasks were included in the ST
intervention. The difficulty of the exercises used in the ST intervention progressed with
difficulty in a similar fashion to the BTT protocol. Again, interested readers can obtain a
detailed description the ST protocol from the corresponding author (Appendix A).
53
3.2.4 Biomechanical testing protocol
To evaluate the influence of BTT and a season AF on an athlete’s knee joint
biomechanics, testing was conducted on two occasions. The first was in weeks 1–7 of the
WAAFL preseason training schedule. The second was conducted in weeks 18–25 of their
respective training interventions.
During biomechanical testing, participants completed a random series of pre-planned and
unplanned straight-run, crossover-cut and sidestep running tasks with their self-selected
preferred stance limb (Figure 3.2).[6–8 12 15 16] A computer monitor displayed a 30 cm
arrow to direct participants to perform the straight-run or change of direction running tasks.
During unplanned running tasks, the direction arrows were triggered by the athlete running
through timing gates situated along the approach pathway. The direction arrow was
signalled by the timing gates when participants were approximately 1.5 m from the force
plate, which corresponded to contralateral leg toe off. For all running tasks, a trial was
considered successful if the average approach velocity of the right anterior superior iliac
spine marker calculated in Vicon workstation (Vicon Peak, Oxford Metrics, UK) was
between 4.5 ms–1 and 5.5 ms–1. A successful change of direction trial also required
participants to contact a black line marked on the running surface 45º relative to global x-
axis of the laboratory with the contralateral leg during cutting manoeuvre (figure 3.2).
Participants were required to complete three successful trials of each running task before
testing was completed. To minimise participant fatigue during the testing period,
participants were restricted to maximum of 30 running trials during testing and were given
at least 60 s of rest between each running task.
54
Figure 3.2
Above: frontal (1) and transverse (2) view of the sidestep sport maneuvers conducted during biomechanical testing. The solid black lines were used as direction cues for participants during change of direction tasks. Below: mid pelvis position (x,y) coordinates 50 frames prior to heel contact (A), at heel contact (B), contralateral leg heel contact (C) and ipsilateral leg mid swing (D) were used to define vectors AB and CD. The cosine of the dot product between vectors AB and CD represents a participants CoD angle during sidestepping.
55
A 12-camera 250 Hz Vicon MX motion capture system (Vicon Peak, Oxford Metrics, UK)
was used to record three-dimensional full-body kinematics.[15 16] Ground reaction forces
(GRF) were synchronised and recorded at 2000 Hz from a single 1.2×1.2 m force plate
(Advanced Mechanical Technology, Watertown, Massachusetts, USA).
Ankle joint centres were defined using anatomical landmarks on the medial and lateral
malleoli. A six-marker pointer was used to digitise the medial and lateral femoral condyles,
with a functional knee axis to define knee joint centres and knee axes orientation.[21] A
functional method was also used to define the hip joint centres.[21] A custom foot
alignment rig was used to measure calcaneous inversion/eversion and foot
abduction/adduction to define the anatomical coordinate system of the foot segment.[21]
Marker trajectories and GRF data were both low pass filtered at 15 Hz using a zero-lag
fourth-order Butterworth filter, which was selected based on a residual analysis[22] and
visual inspection.
3.2.5 Analysis
Spatial–temporal, knee kinematic and knee kinetic variables were analysed during pre-
planned running (PpRun), PpSS and UnSS sport manoeuvres. Spatial–temporal variables
included mean pre-contact (PC) velocity, mean change of direction (CoD) angle and mean
CoD velocity. The PC velocity was calculated as the mean mid-pelvis horizontal velocity
50 ms prior to heel contact. The CoD angle was calculated by taking the cosine dot
product of two vectors representing the position of (a) mid-pelvis 50 frames before heel
contact to (b) heel contact and (c) contralateral leg heel contact to (d) ipsilateral leg mid
swing (Figure 3.2). The CoD velocity was determined from the mid-pelvis resultant velocity
during the first 3/4 of stride. The last quarter of stride was not used as this typically
occurred outside the calibrated motion capture volume of the laboratory.
Knee kinematics and kinetics were calculated within the WA phase of stance (heel contact
to first trough in vertical GRF vector) using custom lower limb kinematic and inverse
dynamic models in Bodybuilder (Vicon Peak, Oxford Metrics, UK).[6–8 12 15 16]
Kinematic variables calculated in this phase included mean knee flexion and knee flexion
range of motion (RoM). Kinetic variables included mean peak externally applied flexion,
56
valgus and internal-rotation knee moments. All knee moments were normalised to each
participant’s total body mass and height.
3.2.6 Statistics
Biomechanical investigators were blinded to each participants training intervention until
final statistics were performed. Only athletes who attended both biomechanical testing
sessions were included in the analysis. All variables described in the analysis section were
assessed using a linear mixed model (α=0.05) in SPSS 17.0.1 (SPSS, IBM Headquarters,
Chicago, Illinois, USA). Factors were time (testing session 1 or 2), training intervention
(BTT or ST) and running task (PpRun, PpSS or UnSS). The number of training sessions
that each athlete participated in-between biomechanical testing sessions was used as a
covariate. An adjusted Sidak post hoc analysis (α=0.05) was used to assess significant
main effects and interactions.
A Pearson’s Correlation (R2), 95% CI and limits of agreement (LoA) for PC velocity, CoD
angle and CoD velocity measures were used to assess the reliability of the UWA
sidestepping protocol between biomechanical testing sessions 1 and 2. Pre-empting the
results, no statistical differences in PC velocity, CoD angle and CoD velocity were
observed between training groups or biomechanical testing sessions, so were grouped
together for the aforementioned correlation and LoA analysis.
3.3 RESULTS No significant differences in knee kinematic variables were observed between training
groups or biomechanical testing sessions for all running tasks (Table 3.1). Mean knee
flexion and knee flexion RoM were significantly different between running tasks (p <
0.001). Post hoc analyses showed that peak knee flexion and knee flexion RoM were
significantly elevated during both sidestepping tasks when compared with PpRun. Only
knee flexion RoM was significantly elevated during UnSS when compared with PpSS.
No significant differences in peak knee flexion, valgus or internal rotation moments were
observed between training groups for all running tasks (Table 3.2). Mean peak knee
flexion moments were significantly different between running tasks (p < 0.001). Post hoc
analysis showed that mean peak knee flexion moments during both sidestepping tasks
were significantly larger than PpRun.
57
Table 3.1
Mean Knee Flexion (deg) Knee Flexion RoM (deg)
PpRun 26.2 ± 4.6 a 19.6 ± 5.2 a
PpSS 29.8 ± 5.4 b 33.0 ± 6.2 b
UnSS 30.0 ± 5.0 b 35.3 ± 6.4 c a,b,c indicates significant Sidak adjusted post hoc difference between independent variables (p < 0.05 ) (n = 34). If two independent variables posses the same letter they are not significantly different from each other.
An interaction between running task and time in peak valgus knee moments was observed
(p = 0.037). Post hoc analysis showed that peak valgus knee moments were significantly
elevated during both sidestepping tasks when compared with PpRun across both testing
sessions. Peak valgus knee moments during UnSS were significantly elevated relative to
PpSS in testing session 2, but not during testing session 1. Peak valgus knee moments
during UnSS significantly increased (p = 0.022) by 31% from testing session 1 (0.48 ± 0.27
Nm.kg-1.m-1) to testing session 2 (0.63 ± 0.40 Nm.kg-1.m-1).
Table 3.2
Flexion (Nm∙kg-1∙m-1)
Valgus (Nm∙kg-1∙m-1)
Int. Rot. (Nm∙kg-1∙m-1)
T1
PpRun 1.44 ± 0.39 a 0.15 ± 0.10 a 0.15 ± 0.09 a
PpSS 2.14 ± 0.55 b 0.37 ± 0.30 b 0.33 ± 0.36 b ,†
UnSS 2.16 ± 0.42 b 0.48 ± 0.27 b ,† 0.20 ± 0.15 a
T2
PpRun 1.34 ± 0.25 a 0.12 ± 0.08 a 0.13 ± 0.08 a
PpSS 2.15 ± 0.42 b 0.35 ± 0.27 b 0.18 ± 0.09 a ,†
UnSS 2.08 ± 0.44 b 0.63 ± 0.40 c ,† 0.15 ± 0.06 a † indicates significant difference over time (p < 0.05) (n = 34). a,b,c indicates significant Sidak adjusted post hoc difference between independent variables (p < 0.05 ) (n = 34). If two independent variables posses the same letter they are not significantly different from each other.
An interaction in peak internal rotation knee moments was observed between running task
and time (p = 0.026). Post hoc analysis showed that in testing session 1, PpSS peak
internal rotation knee moments were significantly elevated relative to both UnSS and
PpRun. In testing session 2, no differences in peak internal rotation knee moments were
observed between running tasks. Peak internal rotation knee moments during PpSS
significantly decreased (p = 0.025) by 45% from testing sessions 1 (0.33 ± 0.36 Nm.kg-1.m-
1) to testing session 2 (0.18 ± 0.09 Nm.kg-1.m-1).
Mean peak flexion, valgus and internal rotation (Int. Rot.) knee moments of both training groups across testing session 1 and 2 for all running tasks.
Mean knee flexion angle and range of motion (RoM) during the weight acceptance phase of stance for all running tasks. BTT and ST groups across both testing sessions 1 and 2 were pooled together.
58
The UWA sidestepping protocol reliability test (Table 3.3) showed a moderate to strong
correlation in CoD angle during both sidestepping tasks (R2 ≥ 0.55). Between testing
sessions 1 and 2, moderate to strong correlations in PC velocity and CoD velocity were
observed during UnSS (R2 ≥ 0.46), while moderate to low correlations were observed
during PpSS and PpRun (R2 ≤ 0.30). This is likely attributed to the use of laser timing
gates to control for velocity during UnSS. The limits of agreement for all velocity
measures were all less than 1.0 ms, which can be considered negligible differences.
Table 3.3
An interaction between running task and time in PC velocity was observed (p = 0.022)
(Table 3.4). Post hoc analysis showed that PC velocity was significantly elevated during
PpSS relative to UnSS during testing session 1, but not for testing session 2. The PC
velocity during PpRun was significantly elevated relative to both sidestepping tasks in both
testing sessions.
Table 3.4
3.4 DISCUSSION Both balance[8] and technique[15] training conducted in controlled laboratory settings
have been shown to be effective in decreasing internal rotation[8] and/or valgus[8 15] knee
moments during both PpSS and UnSS. However, to date, no study has determined if BTT-
implemented adjunct to normal ‘real-world’ preseason and regular season training is
effective in reducing peak knee moments during the WA phase of PpSS and UnSS.
Additionally, it is unknown if an athlete’s knee joint biomechanics change over a playing
season. The major finding of this study was BTT implemented in a ‘real-world’ community-
level training environment did not change an athlete’s laboratory measurements of knee
CoD Angle (R2; 95% CI)
PC Velocity (R2; 95% CI)
CoD Velocity (R2; 95% CI)
CoD Angle (LoA)
PC Velocity (LoA)
CoD Velocity (LoA)
PpRun -- 0.21; 0.13-0.70 -- -- ± 0.9 ms-1
--
PpSS 0.55; 0.54-0.86 0.30; 0.25-0.75 0.29; 0.24-0.74 ± 6.1° ± 0.9 ms-1
± 0.9 ms-1
UnSS 0.69; 0.68-0.92 0.46; 0.45-0.83 0.50; 0.47-0.85 ± 6.8° ± 1.0 ms-1
± 0.6 ms-1
1
CoD Angle (°)
CoD Velocity (m/s)
Testing Session 1 Testing Session 2
PC Velocity (m/s) PC Velocity (m/s)
PpRun 1.0 ± 0.60 a 5.4 ± 1.68
a 5.4 ± 0.50
a 5.3 ± 0.40
a
PpSS 16.0 ± 3.16 b
4.6 ± 0.49 b 5.1 ± 0.50
b 5.1 ± 0.42
b
UnSS 16.0 ± 3.21 b 4.4 ± 0.55
b 4.9 ± 0.48
c 5.0 ± 0.44
b
a,b,c indicates significant Sidak adjusted post hoc difference between independent variables (p < 0.05 ) (n = 34). If
two independent variables posses the same letter they are not significantly different from each other.
Pearson correlation (R2), 95% confidence interval (95% CI) and limits of agreement
(LoA) for change of direction (CoD) angle, pre-contact (PC) velocity and CoD velocity between testing session 1 and 2 for all running tasks.
Mean sidestep CoD angle, CoD velocity and PC velocity for both training groups and across all running tasks. PC velocity was reported for testing sessions 1 and 2.
59
joint biomechanics during either PpSS or UnSS. However, knee moments during both
PpSS and UnSS tasks were found to respond differently over the playing season.
The main finding of this study is that 28 weeks of BTT was not effective in reducing
external knee moments when implemented adjunct to normal ‘real-world’ AF training.
These results do not align with previous literature. Neuromuscular[23] and Plyometric[24]
training conducted in ‘ideal’ training settings were both effective in decreasing valgus knee
moments during double-leg drop-landing sport tasks. A condensed Plyometric based
training protocol implemented adjunct to regular season basketball training was then
shown to be effective in reducing peak extension and valgus knee moments during a
similar double-leg drop-landing task (Y. S. Lee, personal communication,
19 January 2011).[25] Chappell and Limpisvasti26 showed that 10–15 min of
neuromuscular training adjunct to preseason soccer and regular season basketball
training was effective in reducing valgus knee moments during double-leg stop-landing
task. These significant results were in part attributed to two factors, (1) high athlete
compliance,25 26 which was as high as 100%,[26] and (2) low coach/trainer to athlete
ratios during training, which was approximately 3:11,[25] and 2:33.[26] Within this study,
we had low athlete compliance (45%) and a relatively high coach to athlete ratio (1:40).
These may have been two major limiting factors preventing the positive biomechanical
responses observed in laboratory-based studies[8 15] from being transferred to a ‘real-
world’ community-level training setting.
It is possible that modifying the BTT protocol may have improved athlete compliance,
which may have resulted in different biomechanical outcomes following training.
Mykleburst et al.[27] modified their balance training protocol midway between a 2-year
training intervention based on athlete and coach feedback. Following these changes, they
observed reductions in ACL injury rates in the second season of their training protocol. It is
evident that future research should give more attention towards addressing an athlete’s
perceptions of and compliance to biomechanically based ACL injury-prevention
protocols.[28] Of equal importance, a coach’s attitudes and beliefs toward an intended
ACL injury-prevention programme must also be addressed to ensure the intended benefits
are effectively translated to the athlete.[29] Considering the psychological needs of both
athletes and coaches when implementing injury-prevention protocols at the community
level will likely increase athlete compliance[30] and therefore the probability of positive
60
biomechanical outcomes associated with balance[8] and technique[15] training being
transferred to ‘real-world’ community-level training environments.[18 29 31]
Within-season training effects were observed in this study. For example, following a
season of AF, both the BTT and ST groups displayed a 45% reduction in internal-rotation
knee moments with no change to either their valgus or flexion knee moments during
PpSS. This could be due to PpSS being a common sport skill within AF, which is
performed repeatedly by athletes during normal training (2 h×2 days per week) and play (2
h×1 day per week) in both training groups. It is therefore possible that athletes learnt to
adopt techniques during their normal AF training and game play that reduced their interna
lrotation knee moments and ACL injury risk.
Within-season training effects were also observed during UnSS; external valgus knee
moments in both the BTT and ST groups increased by 31% between testing sessions.
These results are supported by previous research, which has shown that following 12
weeks of normal AF football training (control group), valgus knee moments increased by
26% during the WA phase of sidestepping.8 It could be argued that differences in CoD
angle and/or velocity between biomechanical training sessions may have contributed to
these observed increases. However, PC velocity, CoD angle or CoD velocity during UnSS
were all shown to be similar between testing sessions, suggesting that these variables are
not associated with the signify cant increases in valgus knee moments. Results show an
athlete may be at increased risk of ACL injury when performing an unplanned sidestep in
the latter half of a playing season. Additionally, these results support previous literature
and show that unplanned sport tasks are unique factors associated with ACL injury risk.[6
32]
The ecological validity of the biomechanical testing protocol used in this investigation may
have influenced knee loading measurements during biomechanical testing and in turn the
non-significant findings associated with the BTT protocol. For example, all running tasks
were conducted under non-fatigued conditions; athletes were tested in isolation rather
than in a team environment and differences in the running surface between the laboratory
and training environment were apparent. However, we should note that previous training
interventions, using the same biomechanical methods have been sensitive enough to
measure significant changes in valgus knee loading pre to post training.[8 15] We
61
therefore believe our ‘non change’ in knee joint biomechanics following BTT were likely
associated with the training intervention and not the biomechanical testing protocol.
3.5 CONCLUSIONS
BTT in ‘real-world’ training environments, adjunct to normal AF training was not effective in
changing an athlete’s knee joint kinematics or decreasing external knee moments during
the WA phase PpSS or UnSS. Knee joint biomechanics respond to normal AF training
differently during both pre-planned and unplanned sidestepping tasks. When performing
an unplanned sidestepping task in the latter half of a playing season, athletes are at an
increased risk of ACL injury. Both pre-planned and unplanned sidestepping tasks are
therefore recommended during biomechanical testing when assessing the effectiveness of
prophylactic training protocols. Athlete compliance to training and coach to athlete ratios
should be considered when implementing training interventions in ‘real-world’ community-
level training environments.
Acknowledgements
We thank Mr. Kevin Murray and Ms. Laura Firth from the UWA Statistical Consulting
Group for statistical advice. Dr Dara Twomey provided support to the PAFIX study in her
role as the Victorian-based Project Manager.
Competing interest statement
There were no financial or personal relationships with other people or organizations that
could have biased the presented work.
Contributor statement
C.J. Donnelly, B.C. Elliott, T.L.A. Doyle, C.F. Finch, A.R. Dempsey, and D.G. Lloyd have
all made substantial contributions to the following: (1) the conception and design of the
study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or
revising it critically for important intellectual content, (3) and the final approval of the
attached manuscript.
62
Funding statement
This study was part of the Preventing Australian football Injuries through exercise (PAFIX)
study funded by Australian National Health and Medical Research Foundation (ID:
400937), as well as the Western Australian Medical and Health Research Infrastructure
Council. Caroline Finch was supported by an NHMRC Principal Research Fellowship (ID:
565900). The Australian Centre for Research into Injury in Sport and its Prevention
(ACRISP) is one of the International Research Centres for Prevention of Injury and
Protection of Athlete Health supported by the International Olympic Committee (IOC).
Reference list chapter 3
1. Gianotti SM, Marshall SW, Hume PA, et al. Incidence of anterior cruciate ligament injury and other knee ligament injuries: a national population-based study. J Sci Med Sport 2009;12(6):622-7.
2. Janssen KW, Orchard JW, Driscoll TR, et al. High incidence and costs for anterior
cruciate ligament reconstructions performed in Australia from 2003-2004 to 2007-2008: time for an anterior cruciate ligament register by Scandinavian model? Scand J Med Sci Sports 2011 doi: 10.1111/j.1600-0838.2010.01253.x
3. Dunn WR, Spindler KP. Predictors of activity level 2 years after anterior cruciate
ligament reconstruction (ACLR): a Multicentre Orthopaedic Outcomes Network (MOON) ACLR cohort study. Am J Sports Med 2010;38(10):2040-50.
4. Roos H, Ornell M, Gardsell P, et al. Soccer after anterior cruciate ligament injury--an
incompatible combination? A national survey of incidence and risk factors and a 7-year follow-up of 310 players. Acta Orthop Scand 1995;66(2):107-12.
5. Cochrane JL, Lloyd DG, Buttfield A, et al. Characteristics of anterior cruciate ligament
injuries in Australian football. J Sci Med Sport 2007;10(2):96-104. 6. Besier TF, Lloyd DG, Ackland TR, et al. Anticipatory effects on knee joint loading during
running and cutting maneuvers. Med Sci Sports Exerc 2001;33(7):1176-81. 7. Besier TF, Lloyd DG, Cochrane JL, et al. External loading of the knee joint during
running and cutting maneuvers. Med Sci Sports Exerc 2001;33(7):1168-75. 8. Cochrane JL, Lloyd DG, Besier TF, et al. Training affects knee kinematics and kinetics
in cutting maneuvers in sport. Med Sci Sports Exerc 2010;42(8):1535-44. 9. Markolf KL, Burchfield DM, Shapiro MM, et al. Combined knee loading states that
generate high anterior cruciate ligament forces. J Orthop Res 1995;13(6):930-5.
63
10. Cerulli G, Benoit DL, Lamontagne M, et al. In vivo anterior cruciate ligament strain behaviour during a rapid deceleration movement: case report. Knee Surg Sports Traumatol Arthrosc 2003;11(5):307-11.
11. Jindrich DL, Besier TF, Lloyd DG. A hypothesis for the function of braking forces
during running turns. J Biomech 2006;39(9):1611-20. 12. Dempsey AR, Lloyd DG, Elliott BC, et al. The effect of technique change on knee
loads during sidestep cutting. Med Sci Sports Exerc 2007;39(10):1765-73. 13. Wu JL, Hosseini A, Kozanek M, et al. Kinematics of the anterior cruciate ligament
during gait. Am J Sports Med, 2010;38(7):1475-1482. 14. Lloyd DG. Rationale for training programs to reduce anterior cruciate ligament injuries
in Australian football. J Orthop Sports Phys Ther 2001;31(11):645-54. 15. Dempsey AR, Lloyd DG, Elliott BC, et al. Changing sidestep cutting technique reduces
knee valgus loading. Am J Sports Med 2009;37(11):2194-200. 16. Donnelly CJ, Lloyd DG, Elliott BC et al. Optimizing whole-body kinematics to minimize
valgus knee loading during sidestepping: Implications for ACL injury risk. J Biomech 2012; 45:1491-1497.
17. Lloyd DG, Buchanan TS. Strategies of muscular support of varus and valgus isometric
loads at the human knee. J Biomech 2001;34(10):1257-67. 18. Finch C. A new framework for research leading to sports injury prevention. J Sci Med
Sport 2006;9(1-2):3-9. 19. Finch C, Lloyd D, Elliott B. The Preventing Australian football Injuries with Exercise
(PAFIX) Study: a group randomised controlled trial. Injury prevention 2009;15(3):e1 doi: 10.1136/ip.2008.021279.
20. Twomey DM, Finch CF, Doyle TL, et al. Level of agreement between field-based data
collectors in a large scale injury prevention randomised controlled trial. J Sci Med Sport 2011;14(2):121-125.
21. Besier TF, Sturnieks DL, Alderson JA, et al. Repeatability of gait data using a
functional hip joint centre and a mean helical knee axis. J Biomech 2003;36(8):1159-68.
22. Winter D. Motor Control of Human Movement. 3 ed. Hoboken, New Jersey: John Wiley
& Sons, Inc., 2005. p. 49-50. 23. Myer GD, Ford KR, Palumbo JP, et al. Neuromuscular training improves performance
and lower-extremity biomechanics in female athletes. J Strength Cond Res 2005;19(1):51-60.
24. Hewett TE, Stroupe AL, Nance TA, et al. Plyometric training in female athletes.
Decreased impact forces and increased hamstring torques. Am J Sports Med 1996;24(6):765-73.
64
25. Lim BO, Lee YS, Kim JG, An KO, et al. Effects of sports injury prevention training on the biomechanical risk factors of anterior cruciate ligament injury in high school female basketball players. Am J Sports Med 2009;37(9):1728-34.
26. Chappell JD, Limpisvasti O. Effect of a neuromuscular training program on the kinetics
and kinematics of jumping tasks. Am J Sports Med 2008;36(6):1081-6. 27. Myklebust G, Engebretsen L, Braekken IH, et al. Prevention of anterior cruciate
ligament injuries in female team handball players: a prospective intervention study over three seasons. Clin J Sport Med 2003;13(2):71-8.
28. Finch CF, White P, Twomey D, et al. Implementing an exercise-training programme to
prevent lower-limb injuries: Considerations for the development of a randomised controlled trial intervention delivery plan. Br J Sports Med. 2011;45(10):791-796.
29. Twomey D, Finch C, Roediger E, et al. Preventing lower limb injuries: is the latest
evidence being translated into the football field? J Sci Med Sport 2009;12(4):452-6. 30. Deci EI, Ryan RM. Intrinsic motivation and self-determination in human behaviour.
New York: Plenum Press, 1985. p. 11-39, 315-332. 31. Finch CF, Donaldson A. A sports setting matrix for understanding the implementation
context for community sport. Br J Sports Med 2010;44(13):973-8. 32. McLean SG, Borotikar B, Lucey SM. Lower limb muscle pre-motor time measures
during a choice reaction task associate with knee abduction loads during dynamic single leg landings. Clin Biomech 2010;25(6):563-9.
65
CHAPTER 4
CHANGES IN MUSCLE ACTIVATION FOLLOWING BALANCE AND TECHNIQUE
TRAINING AND A SEASON OF AUSTRALIAN FOOTBALL.
Donnelly C.J., Elliott, B.C., Doyle, T.L.A., Finch, C.F., Dempsey, A.R. and Lloyd, D.G.
Changes in muscle activation following balance and technique training and a season of
Austrian Football. Br J Sports Med. [Submitted], December, 2011.
The PhD candidate, Cyril J. Donnelly accounted for 70% of the intellectual property
associated with the final manuscript. Collectively, the remaining authors contributed 30%.
Conflict of Interest: There were no financial or personal relationships with other people or
organizations that could have biased the presented work
66
Abstract
Purpose: Determine if balance and technique training (BTT) implemented adjunct to
normal Australian football (AF) training influences the activation of muscles crossing the
knee during sidestepping. Furthermore, determine if changes in muscle activation are
proportional to changes in knee loading. Methodology: 1,001 amateur AF players
participated in either 28 weeks of BTT or ‘sham’ training (ST) adjunct to their normal
training. Twenty-eight athletes (BTT, n = 12; ST, n = 16) completed biomechanical testing
prior to and following training. Directed co-contraction ratios (DCCR) in three degrees of
freedom and total muscle activation (TMA) were calculated during pre-planned (PpSS)
and unplanned (UnSS) sidestepping. Changes in muscle activation were assessed with
changes in knee loading as described by in chapter 3. Results: BTT did not influence the
activation of the muscles crossing the knee during sidestepping. However, following a
season of AF, significant increases in knee extensor (p = 0.023) and semimembranosus (p
= 0.006) muscle activation were observed during both PpSS and UnSS. Following a
season of AF, TMA was lower during UnSS when compared to PpSS, even in the
presence of significantly (p = 0.022) elevated valgus knee moments. Conclusions: BTT
was not effective in changing the activation of the muscles crossing the knee during
sidestepping when conducted in ‘real-world’ training environments. Following a season of
AF, athletes are better able to support the knee from both frontal and sagittal plane knee
loading during PpSS and UnSS. Elevated valgus knee loading combined with relatively
lower TMA during UnSS following a season of AF suggests an athlete may be at
increased risk of ACL injury when conducting unplanned sports tasks in the second half of
a playing season.
KEYWORDS: MUSCLE ACTIVATION; INJURY PREVENTION; TRAINING; KNEE; ACL
4.1 INTRODUCTION
From chapter 3 of this investigation,[1] two biomechanical approaches can be used to
reduce anterior cruciate ligament (ACL) injury risk during non-contact sporting tasks like
sidestepping. First, decrease the external loading applied to the knee and ACL by
changing an athlete’s technique during the sport task.[2-4] Second, increase the strength
and/or activation of the muscles crossing the knee capable of protecting the knee and ACL
when loading is elevated.[5] Increasing the activation of muscles with moment arms
capable of supporting the knee from applied flexion, valgus and internal rotation knee
67
moments are thought to be appropriate neuromuscular responses,[4,6] as these loading
patterns have been shown to elevate ACL strain.[7] With no single muscle crossing the
knee capable providing support in all three degrees of freedom simultaneously, different
muscle activation strategies can be used to support the knee and reduce ACL injury risk
during dynamic sporting tasks like sidestepping.
When simulating the contact phase of landing in a cadaveric knee model, Hashemi et
al.[8] found that increased quadriceps force in the pre-contact (PC) phase of landing
resulted in lower ACL strain during the impact phase. Reductions in ACL strain were
attributed to the ability of the quadriceps to prevent tibial translations relative to the femur
by both increasing joint stiffness at low knee flexion angles, and producing a posteriorly
directed joint reaction forces past 20°of knee flexion.[8] Due to their line of action,
hamstring muscle force can reduce ACL tension from 15-45º of knee flexion[9] and reduce
ACL strain further when the hamstring is co-contracted with the quadriceps.[10]
Valgus and internal rotation knee moments can be supported with specific knee muscle
activation patterns.[5] Generally, medial knee muscles have moment arms capable of
supporting valgus knee moments[5, 11-13] and considered an appropriate strategy for
supporting the ACL from external valgus knee moments.[5, 11] In summary, appropriate
muscle activation strategies to counter applied flexion, valgus and/or internal rotation knee
moments, during sidestepping include generalised hamstring/quadriceps co-contraction,
superimposed with the increased activation of muscles with flexion, and/or medial moment
arms[14].
Training with unstable bases of support have been shown to elevate muscle activation and
hamstring/quadriceps co-contraction during resistance training.[15] Twenty minutes of in-
season neuromuscular training, which contains balance training components was found to
be effective in increasing medial hamstring muscle activation during sidestepping.[16]
However, external knee loading was not measured, nor a randomized control group or
cross-over experimental design used.[16] It is then unclear if the observed changes in
hamstring muscle activation are in response to changes in knee joint loading[11] or simply
due to normal in-season training. The efficacy of balance training must be tested in ‘real-
world’ settings, and analysed in conjunction with changes in knee loading before it can be
recommended to community level athletes.
68
There were three purposes of this investigation 1) determine if balance and technique
(BTT) implemented adjunct to pre-season and regular season Australian football (AF)
training influenced the activation patterns of the muscles crossing the knee during pre-
planned (PpSS) and unplanned (UnSS) sidestepping. 2) Determine if an athlete’s muscle
activation changed over a normal season of AF and 3) determine if changes in muscle
activation following BTT and/or a season of AF were proportional to changes in knee
loading. We hypothesise that following BTT, total muscle activation and co-contraction
between the knee flexor and extensor muscle groups will increase. We also hypothesise
the relative activation of muscles with medial moment arms relative to lateral moment
arms will increase following BTT.
4.2 METHODS
The methods summarized here are a condensed version of those presented previously in
chapter 3 of this study[1] and in the Preventing Australian football Injuries through
eXercise (PAFIX) study protocol[17] (Appendix A). This study was approved by the
Human Research Ethics Committees at The University of Western Australia (UWA) and
the University of Ballarat.
4.2.1 Participant population – training intervention
Eight amateur level AF clubs (n=1,001 males) volunteered to participate in either 28 weeks
of BTT or a ‘sham’ training (ST) intervention adjunct to their regular season training. All
participants provided their informed, written consent prior to participating in their respective
training interventions.
4.2.2 Participant population – biomechanical testing
Fifty-eight athletes were randomly recruited in weeks -1 to 7 for biomechanical testing.
Thirty-four returned for testing in weeks 18 to 25. Both knee loading and surface
electromyography (sEMG) data was obtained from 28 (48%) participants (BTT, n = 12; ST,
n = 16) (Figure 4.1). All participants provided their informed, written consent prior to
biomechanical testing.
69
Figure 4.1
4.2.3 Training protocol
Each club trained two times per week and played a match once a week over the 28 week
training interventions. Training interventions were conducted as a pre-training warm-up for
20 minutes, twice a week for the first 18 weeks, and then once a week for the remaining
28 weeks. Training sessions were run by two qualified instructors. Instructors were
blinded to 1) which training programs they were overseeing, and 2) the outcome variables
analysed during biomechanical testing were assigned to each of the eight WAAFL clubs.
Instructors recorded player attendance and participation following each training session.
Testing Session 1 (wks -1 to 7) Testing Session 2 (wks 18 to 25)
ST (n = 16) BTT (n = 12) ST (n = 16) BTT (n = 12)
Age (yrs) 21.2 ± 2.7 21.2 ± 3.7 21.9 ± 2.8 21.5 ± 3.1
Height (m) 1.84 ± 0.08 1.86 ± 0.09 1.84 ± 0.08 1.86 ± 0.09
Mass (kg) 81.6 ± 9.9 82.5 ± 10.2 81.4 ± 9.9 82.2 ± 10.6
ST (n = 20)
BTT (n = 14)
n = 58
n = 1,001
ST (n = 16)
BTT (n = 12)
Weeks -1 to 7
Weeks18 to 25
Usable sEMG
Experimental data flow of training intervention and biomechanical testing sessions 1 and 2. BTT and ST numbers were only reported in testing session two as the biomechanists conducting the data collections were blinded to the training intervention codes of each participant until the statistics phase of the analysis. Mean ± standard deviation age, body mass and height were reported for participants who completed both testing session 1 and 2.
70
4.2.4 Biomechanical testing protocol
The first biomechanical testing session was in weeks -1 to 7 of the pre-season training
schedule, and the second in weeks 18 to 25. Each testing session started with an
assessment of each participant’s general athletic ability and lower limb strength.
Assessments included maximum effort isometric hip abduction/adduction torque, isokinetic
eccentric knee flexion and extension torque, counter movement vertical jump (CMJ) height
and a single-leg balance test.
Participants then completed the UWA sidestepping protocol,[3, 6] which consists of a
random series of pre-planned and unplanned straight run, crossover and sidestep sporting
manoeuvres with their self selected preferred leg. Participants were required to complete
three successful trials of each sporting manoeuvre before testing was complete.
To assess isometric hip strength, a belt was positioned around each athlete’s femoral
condyles and then attached in series to a force transducer (Fitness Technology Inc., Skye,
Australia) and a vertically adjustable wall mount. With the knee near full extension, and
the hip at approximately 0° and 15° abduction and flexion respectively, participants were
instructed to maintain a vertical trunk posture while producing tension through the belt
during maximum effort isometric hip adduction and abduction contractions. Peak isometric
hip abduction and adduction torques were calculated by multiplying peak force, by femur
length calculated during subject-specific kinematic modelling. A dynamometer (Biodex
System 3, Biodex Medical Systems, Inc., Shirley, NY) was used to record peak eccentric
knee flexion and extension torque at 300°/s of their preferred sidestepping leg. The joint
angles tested were over each participant’s total passive range of motion. Athletes
performed two maximum effort CMJ’s on a single 1.2×1.2 m force plate, where ground
reaction force (GRF) measures were recorded at 2,000 Hz (Advanced Mechanical
Technology Inc., Watertown, MA.). GRF measures were used with the impulse-
momentum method[18] to calculate peak centre of mass (CoM) vertical displacement.
The CMJ with the largest peak vertical displacement was used for analysis.
Participants were asked to perform two single-leg whole-body balance (WBB) tests with
their preferred sidestepping leg for 30 seconds on a 1.2×1.2 m force plate. With eyes
closed and head tilted back, participants were instructed to bring their contralateral leg to
90º of hip and knee flexion. Participants were allowed to place their contralateral foot on
71
the force plate when they felt unstable, but were asked to regain 90º of hip and knee
flexion once they felt ‘stable’. WBB was scored using both GRF and kinematic data
recorded at 2,000 Hz and 250 Hz respectively. When a participant’s GRF vector deviated
outside an area defined by the kinematic markers of their stance foot, their balance score
increased by one. Mean WBB score was calculated from both tests.
During the running and sidestepping trials, sEMG data was collected using a 16-channel
telemetry system (TeleMyo 2400 G2, Noraxon, Scottsdale, Arizona) at 1,500 Hz with a 16
bit A/D card. Input impedance was >100 M and CMR was >100 dB. Data was
synchronised with kinematic and GRF data in Vicon workstation (Vicon Peak, Oxford
Metrics Ltd., UK). Prior to electrode placement, the skin was prepared by shaving,
exfoliating and then cleaning with alcohol. Biopolar 30 mm disposable surface electrodes
(Cleartrace™ Ag/AgCl, ConMed, Utica, NY), with an inter-electrode distance of 30 mm
were placed over the muscle bellies of eight muscles crossing the knee (tensor fasciae
latae (TFL) semimembranosus (SM), biceps femoris (BF), vastus lateralis (VL), vastus
medialis (VM), rectus femoris (RF), medial gastrocnemius (MG) and lateral gastrocnemius
(LG)). Manual clinical muscle testing was used to ensure excitations of each muscle were
being recorded by the corresponding electrodes.
Using customised software in MatLab (Matlab 7.8, The Math Works, Inc., Natick,
Massachusetts, USA), the sEMG data was processed by first removing any direct current
offsets, then band-pass filtered with a 4th order Butterworth digital filter between 30 and
500 Hz. The signal was then full-wave rectified, then linear enveloped by low-pass filtering
with a zero-lag 4th order Butterworth at 6 Hz. Following linear enveloping, peak muscle
activation from each muscle (n = 8) recorded during pre-planned running (PpRun) was
used to normalize each muscle’s sEMG signal to 100% activation.[6]
4.2.5 Analysis
Muscle activation patterns were assessed using total muscle activation (TMA) and
directed co-contraction ratios (DCCR)(Appendix C).[19] The TMA was calculated by
taking the sum of the normalized activation of all muscles crossing the knee. The TMA of
the hamstrings muscles were also calculated and denoted hamstrings-TMA. The DCCR
were calculated for flexion/extension (F/E) muscle groups, medial/lateral (M/L) muscle
groups and the semimembranosus/biceps femoris (SM/BF). Muscles were grouped
72
according to their ability to produce moments in flexion/extension, varus/valgus and
internal/external rotation (Table 4.1).[5, 6, 13, 20] A DCCR is a ratio between 1 and -1,
providing directionality between agonist muscles (flexor and/or medial moment arms) and
antagonist muscles (extensor and/or lateral moment arms). A DCCR > 0 would indicate
co-contraction is directed towards muscles with flexion and/or medial moment arms, while
a DCCR < 0 is directed towards muscles with extension and/or lateral moment arms. A
DCCR = 0 indicates equal activation of agonist and antagonist muscle groups.
Table 4.1
Flexion Extension Varus Valgus Int. Rotation Ext. Rotation
SM BF MG LG
VM VL RF TFL
SM MG VM
BF LG VL
TFL
SM BF
SM (Semimembranosus), BF (Biceps femoris), MG (Medial Gastrocnemius), LG (Lateral Gastrocnemius), VM (Vastus medialis), VL (Vastus lateralis), RF (Rectus femoris), TFL (Tensor fasciae latae).
Muscle activation variables during sidestepping were calculated in two phases; during
weight acceptance (WA) and pre-contact (PC). WA was the period from initial foot contact
to the first trough after the weight acceptance transient in the vertical GRF vector, while
PC was defined as the period 50 ms prior to WA.[3, 6] Muscle activation variables
calculated were mean TMA, mean hamstrings-TMA, mean F/E DCCR, mean M/L DCCR
and mean SM/BF DCCR.
Mean knee flexion (deg), knee flexion RoM (deg) as well as mean peak knee flexion,
valgus and internal rotation moments (Nm∙kg-1∙m-1) were calculated during WA using
custom kinematic and kinetic models in Bodybuilder (VICON Peak, Oxford Metrics Ltd.,
UK) as described in chapter 3. All knee moments were normalized by dividing by the
product of each participant’s total body mass and height.
4.2.6 Statistics
Biomechanical investigators were blinded to which training intervention each athlete
participated in until final statistics were performed. Only athletes from both biomechanical
testing sessions were included for analysis. All variables described in the analysis section
were assessed using a linear mixed model in SPSS 17.0.1 (SPSS Inc, IBM Headquarters,
Muscles grouped according to ability to produce knee moments during flexion, extension, varus, valgus, internal and external rotation degree-of-freedom from 20 to 50 degrees of knee flexion [4, 5, 12, 20].
73
Chicago, Illinois)(α = 0.05). Factors used were time (testing session 1 or 2), training
intervention (BTT or ST), running task (PpRun, PpSS or UnSS) and phase (PC or WA).
The number of training sessions each athlete participated in between testing sessions was
used as a covariate. An adjusted Sidak post hoc analysis (α = 0.05) was used to assess
significant main effects and interactions.
4.3 RESULTS
Significant differences in TMA, F/E DCCR and M/L DCCR were observed between the PC
and WA phase for all running tasks (p < 0.01) (Table 4.2). Conversely, no differences in
hamstring-TMA or SM/BF DCCR were observed between PC and WA phases for all
running tasks, so data were collapsed into one phase for analyses (Table 4.3).
The TMA was significantly elevated during WA when compared with PC (p < 0.001) and
significantly increased from testing sessions 1 to 2 (p = 0.001) for all running tasks within
both phases (Table 4.2). An interaction between running task and training intervention
was observed for TMA (p = 0.022). Post hoc analysis showed that TMA during
sidestepping tasks were significantly elevated relative to PpRun in both the ST and BTT
groups. TMA was elevated during PpSS relative to UnSS in both training groups, but
significance was only attained in the BTT group (p = 0.008).
An interaction between phase and running task was observed for F/E DCCR (p = 0.016)
(Table 4.2). Post hoc analysis showed F/E DCCR was directed towards muscle with
flexion moment arms during PC and extension moment arms during WA for all running
tasks. During PC, the F/E DCCR was further directed towards flexion during PpRun when
compared to the sidestepping tasks. Furthermore, the F/E DCCR was more directed
towards flexion during PpSS when compared with UnSS. During WA, F/E DCCR was
more directed towards extension during sidestepping tasks when compared with PpRun.
No differences were observed between PpSS and UnSS. F/E DCCR across both phases
and all running tasks became directed more towards muscles with extension moment
arms from testing session 1 to 2 (p = 0.023); meaning the relative activation of the
quadriceps and TFL increased over time during both PC and WA.
74
Table 4.2
During testing session 1, SM/BF DCCR was directed laterally towards the BF, for all
running tasks. Between testing session 1 and 2 SM/BF DCCR significantly changed (p =
0.006) and co-contraction increased (SM/BF DCCR = 0), meaning the relative activation of
the SM increased for all running tasks. No significant differences in SM/BF DCCR were
observed between training groups or running tasks (Table 4.3).
No statistical differences in peak isokinetic flexion or extension knee torques, CMJ height
or WBB score were observed between training groups or testing sessions (Table 4.4).
The ST training group displayed a 29% increase in preferred sidestepping leg peak
isometric hip abduction torque between testing sessions 1 (133 ± 29.2 Nm) and 2 (172 ±
58.8 Nm) (p = 0.016).
Phase: Pre-contact TMA F/E DCCR M/L DCCR
ST BTT
Testing Session 1
PpRun 1.80 ± 0.43 †,a
1.95 ± 0.38 †,a
0.62 ± 0.15 †,a
0.08 ± 0.16 a
PpSS 2.56 ± 0.48 †,b
2.87 ± 0.67 †,b
0.38 ± 0.24 †,b
-0.03 ± 0.19 b
UnSS 2.71 ± 0.94 †,b
2.56 ± 0.81 †,c
0.17 ± 0.39 †,c
-0.09 ± 0.27 b
Testing Session 2
PpRun 2.01 ± 0.43 †,a
2.36 ± 0.61 †,a
0.55 ± 0.21 †,a
0.14 ± 0.15 a
PpSS 3.18 ± 0.93 †,b
3.30 ± 0.70 †,b
0.22 ± 0.33 †,b
-0.06 ± 0.25 b
UnSS 3.10 ± 1.23 †,b
3.01 ± 0.79 †,c
0.11 ± 0.30 †,c
-0.10 ± 0.22 b
Phase: Weight Acceptance TMA F/E DCCR M/L DCCR
ST BTT
Testing Session 1
PpRun 2.61 ± 0.42 †,a
2.84 ± 0.42 †,a
-0.03 ± 0.27 †,a
0.02 ± 0.17 a
PpSS 3.68 ± 0.58 †,b
3.82 ± 0.86 †,b
-0.27 ± 0.26 †,b
-0.08 ± 0.20 b
UnSS 3.69 ± 1.01 †,b
3.46 ± 0.68 †,c
-0.29 ± 0.23 †,b
-0.08 ± 0.20 b
Testing Session 2
PpRun 2.77 ± 0.61 †,a
3.27 ± 0.75 †,a
-0.03 ± 0.24 †,a
0.04 ± 0.17 a
PpSS 4.39 ± 0.94 †,b
4.29 ± 0.74 †,b
-0.38 ± 0.19 †,b
-0.16 ± 0.23 b
UnSS 4.09 ± 1.22 †,b
3.78 ± 0.71 †,c
-0.39 ± 0.23 †,b
-0.11 ± 0.27 b
† indicates significant difference over time (p < 0.05) (n = 28).
a,b,c indicates significant Sidak adjusted post hoc difference between independent variables (p < 0.05 ) (n = 28).
If two independent variables posses the same letter they are not significantly different from each other.
TMA and DCCR of the muscles crossing the knee with flexion/extension (F/E) and medial/lateral (M/L) moment arms. Data is presented for testing sessions 1 and 2, during both the pre-contact and weight acceptance phases of running and sidestepping. ST and BTT groups were pooled together unless an interaction was observed. DCCR > 0 co-contraction is directed towards muscles with flexion and/or medial moment arms. DCCR < 0 co-contraction is directed towards muscles with extension and/or lateral moment arms. DCCR = 0 maximal co-contraction.
75
Table 4.3
Table 4.4
4.4 DISCUSSION
The major finding of this study was that BTT implemented adjunct to pre-season and
regular season AF training did not change the activation patterns or strength of the
muscles crossing the knee during either PpSS or UnSS. However, results did show that
following normal AF training, the TMA of the muscles crossing the knee increased, while
DCCR were directed towards muscles with extensor moment arms and the SM, which can
produce flexion, valgus and internal rotation moments. These changes in muscle
activation are effective in countering applied flexion and valgus knee moments
respectively. Without considering changes in external knee loading, one could conclude
that following normal AF training, ACL injury risk was reduced during both PpSS and
UnSS.[15] However, when analysing muscle activation changes in conjunction with the
Phase: Pre-Contact &
Weight Acceptance
Hamstrings-TMA SM/BF DCCR
Testing Session 1
PpRun 0.94 ± 0.33 -0.16 ± 0.24 †
PpSS 1.11 ± 0.42 -0.14 ± 0.28 †
UnSS 0.90 ± 0.36 -0.11 ± 0.32 †
Testing Session 2
PpRun 1.01 ± 0.34 0.00 ± 0.26 †
PpSS 1.07 ± 0.38 0.00 ± 0.31 †
UnSS 0.91 ± 0.33 0.01 ± 0.34 † † indicates significant difference over time (p < 0.05) (n = 28).
Testing Session 1 Testing Session 2
ST BTT ST BTT
Balance Score (#) 10 ± 5.2 8 ± 2.4 9 ± 5.7 8 ± 3.3
CMJ Ht (m) 0.39 ± 0.05 0.42 ± 0.07 0.41 ± 0.06 0.41 ± 0.05
Peak Knee Ext (Nm) 266 ± 71.3 290 ± 76.3 248 ± 78.2 267 ± 62.6
Peak Knee Flex (Nm) 182 ± 44.6 178 ± 44.2 177 ± 48.5 206 ± 52.7
*Pref. Hip ABduction (Nm) 133 ± 29.2 † 198 ± 57.9 172 ± 58.8
† 187 ± 68.4
*Pref. Hip ADduction (Nm) 159 ± 46.8 188 ± 60.7 180 ± 60.4 187 ± 64.1
Non *Pref. Hip ABduction (Nm) 147 ± 31.8 180 ± 71.3 158 ± 53.0 197 ± 78.6
Non *Pref. Hip ADduction (Nm) 161 ± 39.6 184 ± 63.2 178 ± 50.4 189 ± 68.6 † indicates significant difference over time (p < 0.05) (n = 28).
*Note: Pref. means preferred sidestepping leg.
Hamstring-TMA and DCCR of the semimembranosus/biceps femoris (SM/BF) muscles. Data is presented for testing sessions 1 and 2, however the ST and BTT groups as well as the data during the pre-contact and weight acceptance phases of running and sidestepping were pooled.
Mean hip torque, knee torque, CMJ height and full body balance score measures for the ST and BTT test groups between testing sessions 1 and 2.
76
significant increases in peak valgus knee moments during UnSS,[1] the aforementioned
conclusions would be considered inappropriate. When analysing changes in muscle
activation and knee loading together, results show muscle activation patterns following a
season of AF may be better suited to protect the knee and ACL from external knee loading
during PpSS when compared with UnSS.
Following a season of AF training, TMA and PC quadriceps muscle activation both
significantly increased during PpSS and UnSS. It should also be noted, that peak flexion
knee moments[1] and knee joint strength measures remained unchanged. Sidestepping
kinematic data presented in part 1[1] shows that during WA, mean knee flexion was 30º
and knee flexion range increased by 33-35º. Therefore, during WA, the quadriceps would
be contracting eccentrically past 20º of knee flexion. Previous research has shown that
during the simulated impact phase of landing, elevated eccentric quadriceps force past 20°
of knee flexion, decreases ACL strain.[8] Experimental studies have also shown that the
quadriceps are capable of supporting the knee against both varus and valgus knee
moments.[5, 7] Following a season of AF, increases in TMA and PC quadriceps muscle
activation likely reduces an athlete’s risk of ACL injury during UnSS and PpSS,[ 5, 7, 8]
specifically in the flexion/extension plane of motion.
Following a season of AF, the activation of the SM relative to the BF increased during both
PpSS and UnSS. Zebis et al.[16] reported similar findings (Table 4.5), showing significant
increases in SM activation during PpSS following neuromuscular training adjunct to
regular season soccer and handball training. The S/M DCCR calculated from data
presented in Zebis et al.[16] were compared to our results and showed SM/BF DCCR
became more directed towards the SM over the playing season. It should be noted that
Zebis et al.[16] did not directly compare their results to an experimental control group and
attributed muscle activation changes to their in-season neuromuscular training program.
In our study, which did compare results to a control group, no differences in muscle
activation were observed between the ST (control) and BTT group. These results suggest
that normal AF training alone likely increased SM activation and reduced ACL injury risk
(support knee from external valgus moments).
77
Table 4.5
Following a season of AF, TMA was lower during UnSS when compared with AnSS even
though valgus knee moments were significantly higher.[1] The relative differences in PC
TMA activation between UnSS and PpSS (ST 6%, BTT -12%) in testing session 1 did not
correspond to the 30% relative difference in valgus knee moments observed during WA.[1]
In testing session 2, the relative difference in TMA activation between UnSS and PpSS
remained the same (ST -3%, BTT -10%), however the relative difference in valgus knee
moments increased by approximately 80% (0.15 Nm.kg-1.m-1).[1] With minimal changes in
muscle strength between testing session 1 and 2, this suggest that following a season of
AF, the muscles crossing the knee were less capable of supporting the knee from valgus
knee moments during UnSS. It is acknowledged that muscle activation is only an
approximation of muscle force.[13, 21] Nevertheless, when knee loading and muscle
activation are analysed together, results suggest that an athlete may be at increased risk
of ACL injury when conducting unplanned sports tasks in the latter half of a playing
season.
4.5 CONCLUSIONS
BTT implemented in ‘real-world’ training environments, adjunct to normal AF training was
not effective in changing the activation of the muscles crossing the knee during PpSS or
UnSS. Following a normal season of AF, knee extensor and SM muscle activation
increased and are better able to support the knee frontal and sagittal plane knee loading
during PpSS and UnSS. Elevated valgus knee loading combined with relatively low TMA
during UnSS following a season of AF suggests an athlete may be at increased risk of
ACL injury when conducting unplanned sports tasks in the second half of a playing
season.
Phase: Pre-Contact TMA F/E DCCR M/L DCCR Hamstrings-
TMA
SM/BF DCCR
Testing
Session 1
Zebis et al.
2008 2.76 -0.29 0.03 0.69 -0.14
Current
Study 2.56 - 2.87 0.38 -0.03 1.11 -0.14
Testing
Session 2
Zebis et al.
2008 2.89 -0.28 0.12 0.76 0.27
Current
Study 3.18 -3.30 0.22 -0.06 1.07 0.00
1
Relevant TMA and DCCR were calculated during PpSS before and after neuromuscular training from data presented by Zebis et al.[16]. The TFL and MG muscles were not recorded by Zebis et al.[16], so were not used to calculate TMA or the DCCR. It should also be noted that the pre-contact phase in Zebis et al.[16] was 10 ms prior stance foot contact, while in this study it was 50 ms.
78
Acknowledgements
We thank Mr. Kevin Murray and Ms. Laura Firth from the UWA Statistical Consulting
Group for statistical advice. Dr Dara Twomey provided support to the PAFIX study in her
role as the Victorian-based Project Manager.
Competing interest statement
There were no financial or personal relationships with other people or organizations that
could have biased the presented work.
Contributor statement
C.J. Donnelly, B.C. Elliott, T.L.A. Doyle, C.F. Finch, A.R. Dempsey, and D.G. Lloyd have
all made substantial contributions to the following: (1) the conception and design of the
study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or
revising it critically for important intellectual content, (3) and the final approval of the
attached manuscript.
Funding statement
This study was part of the Preventing Australian football Injuries through exercise (PAFIX)
study funded by Australian National Health and Medical Research Foundation (ID:
400937), as well as the Western Australian Medical and Health Research Infrastructure
Council. Caroline Finch was supported by an NHMRC Principal Research Fellowship (ID:
565900). The Australian Centre for Research into Injury in Sport and its Prevention
(ACRISP) is one of the International Research Centres for Prevention of Injury and
Protection of Athlete Health supported by the International Olympic Committee (IOC).
79
Reference list chapter 4
1. Donnelly C.J., Elliott, B.C., Doyle, T.L.A., Finch, C.F., Dempsey, A.R. and Lloyd, D.G.
(2012). Changes in knee joint biomechanics during sidestepping following balance and technique training. Br J Sports Med. doi: 10.1136/bjsports-2011-090829.
2. Dempsey AR, Lloyd DG, Elliott BC, Steele JR, Munro BJ. Changing sidestep cutting
technique reduces knee valgus loading. Am J Sports Med 2009;37(11):2194-200. 3. Dempsey AR, Lloyd DG, Elliott BC, Steele JR, Munro BJ, Russo KA. The effect of
technique change on knee loads during sidestep cutting. Med Sci Sports Exerc 2007;39(10):1765-73.
4. Lloyd DG. Rationale for training programs to reduce anterior cruciate ligament injuries in
Australian football. J Orthop Sports Phys Ther 2001;31(11):645-54; discussion 61 5. Lloyd DG, Buchanan TS. Strategies of muscular support of varus and valgus isometric
loads at the human knee. J Biomech 2001;34(10):1257-67. 6. 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. 7. Markolf KL, Burchfield DM, Shapiro MM, Shepard MF, Finerman GA, Slauterbeck JL.
Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res 1995;13(6):930-5.
8. Hashemi J, Breighner R, Jang TH, Chandrashekar N, Ekwaro-Osire S, Slauterbeck JR.
Increasing pre-activation of the quadriceps muscle protects the anterior cruciate ligament during the landing phase of a jump: an in vitro simulation. Knee 2010;17(3):235-41.
9. More RC, Karras BT, Neiman R, Fritschy D, Woo SL, Daniel DM. Hamstrings--an
anterior cruciate ligament protagonist. An in vitro study. Am J Sports Med 1993;21(2):231-7.
10. Withrow TJ, Huston LJ, Wojtys EM, Ashton-Miller JA. Effect of varying hamstring
tension on anterior cruciate ligament strain during in vitro impulsive knee flexion and compression loading. J Bone Joint Surg Am 2008;90(4):815-23.
11. Buchanan TS, Lloyd DG. Muscle activation at the human knee during isometric flexion-
extension and varus-valgus loads. J Orthop Res 1997;15(1):11-7. 12. Lloyd DG, Buchanan TS. A model of load sharing between muscles and soft tissues at
the human knee during static tasks. J Biomech Eng 1996;118(3):367-76. 13. Lloyd DG, Buchanan TS, Besier TF. Neuromuscular biomechanical modelling to
understand knee ligament loading. Med Sci Sports Exerc 2005;37(11):1939-47. 14. Donnelly, C.J., Elliott, B.C., Ackland T.R., Doyle T.L.A, Besier T.F., Finch, C.F.,
Cochrane, J.L., Dempsey A.R., and Lloyd, D.G. (2012). An anterior cruciate
80
ligament injury prevention framework: Incorporating the recent evidence. Res Sports Med. doi:10.1080/15438627.2012.680989.
15. Behm DG, Anderson K, Curnew RS. Muscle force and activation under stable and
unstable conditions. J Strength Cond Res 2002;16(3):416-22. 16. Zebis MK, Bencke J, Andersen LL, et al. The effects of neuromuscular training on
knee joint motor control during sidecutting in female elite soccer and handball players. Clin J Sport Med 2008;18(4):329-37.
17. Finch C, Lloyd D, Elliott B. The Preventing Australian football Injuries with Exercise
(PAFIX) Study: a group randomised controlled trial. Injury prevention. 2009;15(3):e1 doi: 10.1136/ip.2008.021279.
18. Lithorne NP. Analysis of standing vertical jumps using a force platform. Am J Phys
2001;69(11):1198-204. 19. Heiden TL, Lloyd DG, Ackland TR. Knee joint kinematics, kinetics and muscle co-
contraction in knee osteoarthritis patient gait. Clin Biomech 2009;24(10):833-41. 20. Buford WL, Jr., Ivey FM, Jr., Nakamura T, Patterson RM, Nguyen DK. Internal/external
rotation moment arms of muscles at the knee: moment arms for the normal knee and the ACL-deficient knee. Knee 2001;8(4):293-303.
21. Lloyd DG, Besier TF. An EMG-driven musculoskeletal model to estimate muscle
forces and knee joint moments in vivo. J Biomech 2003;36(6):765-76.
81
CHAPTER 5
AN OPEN-SOURCE COMPUTATIONAL METHOD TO OPTIMISE SIMULATED
HUMAN MOTION TO REDUCE VALGUS KNEE LOADING DURING SIDESTEPPING
AND SINGLE-LEG LANDING.
The PhD candidate, Cyril J. Donnelly accounted for 45% of the intellectual property
associated with this chapter. Jeffery A. Reinbolt contributed 45%. Collectively, Bruce
Elliott and David G. Lloyd contributed 10%.
Conflict of Interest: There were no financial or personal relationships with other people or
organizations that could have biased the presented work
82
Abstract
Externally applied valgus knee moments during the weight acceptance (WA) phase of
sidestepping and single-leg landing (SLL) have been identified as a risk factor associated
with non-contact ACL injuries. Though a model for the aetiology of ACL injuries has been
identified, the kinematics that elevated valgus knee loading are not well understood.
Using an outer-level optimisation technique and the Residual Reduction Algorithm (RRA)
in OpenSim, a method to optimise human motion to minimise peak valgus knee loading
and subsequent ACL injury risk during unplanned sidestepping (UnSS) and SLL was
developed. This approach enabled a dynamically consistent simulation of the stance
phase of straight-line over-ground running to be created (peak RMS kinematic errors <
4.0°; residual errors < 0.3N and 0.4Nm). Further, dynamically consistent simulations of
the WA phase of UnSS (peak RMS kinematic errors < 3.0°; residual errors 2N and 1Nm)
and SLL (peak RMS kinematic errors < 4.0°; residual errors 1N and 1Nm) were also
created. Finally, by executing RRA again, the kinematics of the UnSS and SLL trials were
optimised to minimise valgus knee loading. Pre-to-post kinematic optimisation, peak
valgus knee torques were reduced by 50% (77.9 Nm) during UnSS and 26% (23.3 Nm)
during SLL. The kinematic changes associated with the reductions in peak valgus knee
torque during UnSS were elevated trunk rotation (2.9°), right shoulder adduction (15.7°),
left shoulder flexion (4.1°) and right hip abduction (3.1°) (Stance limb, right leg). The
kinematic changes during SLL were elevated left hip (7.8º) and knee (19.3º) extension
(Stance limb, right leg). An outer-level optimisation technique with the RAA in OpenSim
can be used to identify causal links between an individual’s whole-body kinematics and
valgus knee torque during both UnSS and SLL sport tasks.
5.1 INTRODUCTION
Anterior cruciate ligament (ACL) injuries are severe sport injuries, with approximately 1.15
to 1.3 professional athletes per team per year sustaining a rupture to their ACL during play
(Caraffa et al., 1996; Orchard and Seward, 2009). Coupled with high health care costs to
treat the injury (Gianotti et al., 2009; Janssen et al., 2011), athletes are at increased risk of
early retirement (Ekstand et al., 1990; Dunn and Spindler, 2010; Roos et al., 1995) and
developing radiographic diagnosed knee osteoarthritis (OA) 10-15 years following a
reconstrction if the ACL injury is accompanied by a meniscal tear (Oiestad et al., 2009).
83
The vast majority of non-contact ACL injuries occur during either sidestepping or single-leg
(SLL) landing sports tasks (Cochrane et al., 2007; Koga et al., 2010). Externally applied
valgus knee moments during the weight acceptance phase (WA) of sidestepping and SLL
have been identified as an ACL injury risk factor (Donnelly et al., 2012; Markolf et al.,
1995; McLean et al., 2004; McLean et al., 2008; Quatman et al., 2011; Shin et al., 2011;
Withrow et al., 2006). Although an ACL injury model has been identified, the kinematics
causing high valgus knee loading are not well understood. It is by identifying these casual
relationships that more targeted and therefore effective ACL injury prevention training
protocols can be developed, and in turn reduce ACL injury rates in community level
athletic populations (Finch 2006).
Methods to optimise full-body kinematics to reduce peak knee loading have been reported
by Fregly et al. (2007). This was performed using a patient-specific musculoskeletal
model, the Newton-Euler equations of motion and optimisation computational methods in
the modelling framework Autolev (Online Dynamics, Inc., Sunnyvale, CA). Autolev was
used to alter the kinematics of a male with knee OA to reduce peak knee varus moments
during the stance phase of gait. These methods represent a first step towards
understanding the casual relationships between the complex, multi-body, dynamics of
human gait and joint loading. However, these computational methods are limited in two
ways:
1) The high level of expertise in computational engineering prevents these methods
from being used by the general biomechanics community.
2) The musculoskeletal model and computational software is not open-source, limiting
these methods to laboratories with proprietary licenses to Autolev.
The open-source musculoskeletal modelling software OpenSim (simtk.org, Stanford, CA)
and the optimisation computational tool, the Residual Reduction Algorithm (RRA) can be
used to bridge these gaps. OpenSim allows users to created forward simulations of
human motion from experimental motion capture data. The RRA is a feedback control
method within this forward simulation process that allows users to generate a dynamically
consistent simulation from a set of actuator forces (i.e. joint torques) to track a desired set
of generalised coordinates (i.e. joint angles) with dynamic consistency to the
experimentally recorded ground reaction force (GRF) measures (Delp et al., 2007; Thelen
and Anderson, 2006). Using the RRA as an interface between a dynamic simulation’s
peak joint torques and joint angles, the ability to optimise the kinematics of a simulation to
84
minimise valgus knee loading during sidestepping and SLL is indeed possible. Therefore,
the aims of this study were to develop a simplified method to create dynamically consistent
simulations of human motion. Specifically, we wanted to show in principle that the open-
source musculoskeletal software OpenSim and RRA can be used to optimise the
kinematics of a simulation to minimise peak valgus knee torques during the WA phase of
sidestepping and SLL.
5.2 METHODS
The methods consist of three phases: 1) experimental motion data collection; 2)
development of an outer-level optimisation method to simplify the process associated with
creating a dynamic simulation of human motion; and 3) apply these methods and the RRA
to optimise a simulation’s kinematics to minimise valgus knee loading during UnSS and
SLL.
5.2.1 Experimental data collection
A single male (height 1.74 m; mass 70.2 kg) was randomly chosen from a larger cohort of
34 Western Australian Amateur Football (WAAFL) players. The participant used in this
study gave their informed consent prior to data collection. Ethics was approved by the
Human Research Ethics Committees at The University of Western Australia (UWA) and
the University of Ballarat. Further details associated with the WAAFL cohort can be found
in chapter 3.
Three-dimensional full-body kinematics and GRF were collected during a straight-line
overground-running trial, an unplanned sidestepping (UnSS) trial and a SLL trial. For all
three sports tasks, the participant chose their right leg as their preferred stance limb. The
running and UnSS trials were performed as described in chapter 3. Further, the SLL
procedure as described by Dempsey et al., (2012) (Figure 5.1), required the participant to
run into the laboratory at 5 ms-1, where an Australian football was suspended
approximately 3 m from the floor of the laboratory directly over a single 1.2x1.2m force
platform (Advanced Mechanical Technology Inc., Watertown, MA.). Participants were
instructed to jump from their right leg and while in flight, the Australian football was
randomly swung medially or laterally to the participants approach direction. After the
participant had successfully made contact with the football in flight, they were instructed to
contact the force platform with the same right leg they jumped from. The SLL trial where
85
the Australian football was swung laterally or away from the right leg was used for further
analysis. Further details associated with the overground-running and sidestepping trials
can be found in chapter 3.
A 12-camera 250 Hz VICON MX motion capture system (VICON Peak, Oxford Metrics
Ltd., UK) recorded 3D full-body kinematics (Dempsey et al., 2007). GRF were
synchronously recorded from the force platform at 2,000 Hz. All kinematic and GRF data
were low pass filtered with at the same cut-off frequencies by a zero-lag 4th order
Butterworth digital filter in Workstation (Vicon Peak, Oxford Metrics Ltd., UK). The running
and UnSS data were filtered at 15 Hz, while the SLL data was filtered at 20 Hz. Cut-off
frequencies were selected based on a residual analysis (Winter, 2005) and visual
inspection. Applying the same filter and cut-off frequency to the motion and GRF data has
been shown to reduce knee joint kinetic artefacts during inverse dynamics (Bisseling and
Hof, 2006).
Functional knee and hip joint methods (Besier et al., 2003) were used to calculate subject-
specific joint centres and axes, employing a custom biomechanical model in Matlab
(Matlab 7.8, The Math Works, Inc., Natick, Massachusetts, USA) and Vicon Bodybuilder
(Besier et al, 2003; Dempsey et al., 2007; 2012). Joint centres, marker trajectories
(Appendix E) and GRF (Appendix F) data were then exported into OpenSim 1.9.1.
In OpenSim 1.9.1 a 14 segment, 37 degree-of-freedom (DoF) rigid-linked skeletal model
driven by 37 ideal torque actuators formed the foundation of the three simulations (See
Figure 6.2 in chapter 6)(Appendix D). For clarity, joint torques, not muscles were used to
drive each simulation. Twenty-nine of the model’s DoF have been described previously
(Hamner et al., 2010), to which we added 2 DoF wrist joints (flexion/extension and
ulnar/radial deviation) and 3 DoF knee joints (flexion/extension, internal/external rotation,
and varus/valgus). Internal/external rotation and the varus/valgus DoF of the knee were
modelled as universal joints, with the same centre of rotation. The knee centre of rotation
moved with the flexion/extension DoF, which was modelled as a planar joint, allowing the
tibia to translate relative to the femur as a function of knee flexion angle (Delp et al., 1990).
Segment lengths were scaled to the participant’s subject-specific joint centre positions,
where segment masses and inertial properties were scaled to the participant’s total body
mass in OpenSim. Inverse kinematics (IK) was used in OpenSim to calculate each
model’s generalised coordinates (i.e. joint angles) during the WA phase of UnSS and SLL.
86
Frontal view of the SLL procedure. Frame 1: participant jumps with preferred jumping leg. Frame 1-4: the ball is swung laterally away from their preferred jumping leg, while the participant is in the flight phase. Frame 8 participant lands with preferred jumping leg on a 1.2x1.2m force platform.
Figure 5.1
5.2.2 Dynamically consistent simulation
Following IK, the generalised coordinates and experimental GRF measures of over-ground
straight-line running were used to develop a simplified, user friendly method to create a
dynamically consistent simulation of human motion (Figure 5.2) (Appendix G). To
adequately summarize this process a brief overview of the Newton-Euler equations of
motion and their use in creating a dynamic simulation of motion is presented (1):
FqGqqCτqMq
)()()(1 ,
1 2
4
3
5
7
6
8 9
(1)
87
where q is the generalised coordinate accelerations due to joint torques, τ , coriolis and
centrifugal forces, ),( qqC , as a function of generalised coordinates, q , and their
velocities, q , gravity, )(qG , and external forces (GRF) applied to the model, F , where
1)(
qM is the inverse of the mass matrix. As observed within the Newton-Euler
equations of motion, a relationship exists between the joint angles ( q ) and joint torques
( τ ) to create a dynamic simulation. The RRA utilizes this relationship by manipulating the
angular accelerations of a skeletal model on a frame by frame basis to create a torque
driven simulation that tracks the experimentally recorded GRF with dynamic consistency.
Inconsistencies between a model’s dynamics and experimental GRF measures (∑Fmodel ≠
GRF) called residual forces are present in all biomechanical models, represent errors and
assumptions in the modelling process (i.e. joint centre and inertial estimates)(Delp et al.,
2007). OpenSim addresses this issue by creating a 6 DoF joint between the pelvis and
ground, holding residual forces and moments not solved for during traditional ID, satisfying
Newton’s second law (∑Fmodel + ∑Fresiduals = GRF). The goal of the RRA is to produce a set
of actuator forces (i.e. joint torques) to generate joint motions that track a desired set of
generalised coordinates, while minimising the model’s residual forces and moments (Delp
et al., 2007; Thelen and Anderson, 2006). The result is a simulation that tracks the
experimentally recorded GRF with dynamic consistency.
The first step of RRA optimises the position of the trunk centre of mass (CoM) to reduce
mean residual force and moment offsets. The second step reduces these residuals further
by slightly adjusting the model’s generalised coordinates by minimising the sum of three
components (2): 1) the weighted ( ) squared errors between the experimental (
)
and simulated ( ) accelerations for each DoF of the skeletal model; 2) the
squared residual forces/torques ( ) (n = 6) proportional to their driving excitation ( )
normalized by their maximum residual forces/torques ( ); and 3) the squared joint
torques ( ) for each DoF ( ) in the model, proportional to their driving excitation ( )
normalized by their maximum torques ( ).
88
(2)
The RRA establishes a set of excitations ( ), which can attain values between 0 and
1 that scale the six residual forces/torques ( ) and joint torques ( in each simulation.
Using numerical integration (i.e. ), the model’s dynamics are generated at
each time step. A time varying set of generalised joint coordinates ( ), residual
forces/torques ( ), and joint torques ( are output, producing a dynamically consistent
torque-driven simulation of motion. During RRA, the orientation of the GRF vector is held
relative to the stance foot’s CoM. Thus the location of the centre of pressure is always in
the same relative orientation to the stance foot’s CoM during the RRA process.
The results of RRA depended on what values the user chooses for the 74 maximal joint
torques, maximal pelvic residual forces/torques and kinematic weightings
6+ . For clarity, this means a researcher’s intuition is used to choose the input
parameters for a forward simulation to be created. To produce the best possible
dynamically consistent simulation, we developed an outer-level optimisation method to
help choose these input parameters rather than relying on a researcher’s intuition alone.
The 74 ( ) user-defined input parameters for our 37 DoF full-body model were
reduced to three with the development of an outer-level cost function. The outer-level cost
function (3) was designed to minimise the squared error between the experimental and
simulated kinematics
, simulated residual errors (i.e.
) and simulated joint torques (i.e. ), over time frames. User-defined
uniform weightings were placed on the experimental kinematics of the pelvis ( ),
the remaining kinematics of the model ( ) and six pelvic residuals ( ), which were
1,000, 500 and 500 respectively.
89
(3)
The design variables adjusted in the outer-level optimisation were the set of acceleration
weightings ( ), maximal residual forces/torques ( ) and maximal joint torques (
)
defining the input parameters for RRA. These parameters were used throughout the
inner-level optimisation (RRA) to generate a dynamic consistent simulation that closely
tracked the experimentally recorded kinematics, with minimised residual error.
Figure 5.2
The subject in this study was a male WAAF player. (a) Movement analysis data, including full body, three-dimensional marker trajectories and GRF, were collected during overground straight-line running. (b) A dynamic simulation of the subject was created using a three-step process: 1) a musculoskeletal model with 37 degrees of freedom driven by 37 actuators was scaled to the participant’s joint centres and total body mass; 2) inverse kinematics determined values of the model’s generalized coordinates from the experimentally recorded kinematic data; and 3) RRA was used to produce an optimal set of excitations that produced a dynamically consistent simulation (Equation 2). Note: an outer-level optimization (Equation 3) determined input parameters for the inner-level optimization (RRA) to generate the dynamically consistent simulation.
90
We used this outer-level optimisation method to create a simulation of the stance phase
(heel contact to toe-off) of over-ground running. We then compared the kinematic tracking
errors, residual forces/torques and joint torques created using the outer-level optimisation
method with those produce by a researcher’s intuition, which is normal practice.
5.2.3 Minimisation of valgus knee loading
Following the creation of dynamically consistent simulations of UnSS and SLL (by outer-
level optimisation), RRA was again applied to optimise the simulation’s kinematics to
minimise peak valgus knee moments. This was accomplished by reducing the maximal
joint torque associated with the knee’s V/V DoF ( ) and RRA re-run using the same
maximal torques, maximal residual forces/torques and kinematic weightings solved for
using the external optimisation method along with the experimental GRF measures (4).
Kinematic constraints were placed on the stance foot to restrict foot translations during the
RRA process.
(4)
Pre-to-post kinematic optimisation, selected kinematic and kinetic variables were analysed
during WA. The mean angular differences for all 37 DoF pre-to-post kinematic
optimisation were calculated in 20% intervals over WA, and kinematic maps created for
the UnSS and SLL simulations (See Figure 6.3 in Chapter 6)(Appendix I). Each map
represented the absolute change in joint angles pre-to-post kinematic optimisation, for
each DoF within the skeletal model (n = 37), in each of the five time intervals within WA.
The mean difference of all joints across all time points during WA was then calculated.
Any joint with an angular kinematic change greater than 2σ above the mean was defined
as a critical joint coordinate and identified as a kinematic change that most influenced the
observed changes in peak valgus knee moments pre-to-post kinematic optimisation.
91
5.3 RESULTS
With only three user defined input parameters (i.e. , and ), the outer-level
optimisation method produced a dynamically consistent simulation similar to the
experimentally recorded motion (Figure 5.3). The running simulation tracked the
experimental data with the largest RMS kinematic error being 3.9º, while RMS residual
errors were all below 0.3N and 0.4Nm.
Figure 5.3
Largest differences ordered by decreasing magnitude for (a) kinematic errors (accelerations integrated twice), (b) residual forces/torques, and (c) joint torques resulting from simulations generated using RRA as defined by a typical users intuition (blue, before) and then by the outer-level optimization method (red, after). Also displayed are 10 of the 74 input parameters chosen by a typical user’s intuition (blue, before) and the outer-level optimization method (red, after). These input parameters include kinematic tracking weights (d), maximum residual forces/torques (e), and (f) maximum joint torques.
92
When comparing the kinematic and residual errors of the running simulation when using
the outer-level optimisation method and a typical user’s intuition the outer-level
optimisation method produced kinematic errors up to 3.1º larger than the user’s intuition.
However, when using the outer-level optimisation method, the RMS residual errors were
up to 151N and 23Nm lower than a typical user’s intuition (Figure 5.3).
Using the outer-level optimisation method, the RMS kinematic errors of the simulation
representing the WA phase of UnSS were less than 3.0°, while residuals were below 2N
and 1Nm. The RMS kinematic errors of the simulation representing the WA phase of SLL
were less than 4.0°, while residuals were below 1N and 1Nm.
The maximum allowable valgus joint torque (
) in the dynamically consistent
simulations of UnSS and SLL were reduced by 50% (77.9 Nm) and 26% (23.3 Nm)
respectively (Figure 5.4). During UnSS, both peak flexion and internal rotation knee
moments increased by 5% (13.7 Nm) and 10% (0.6 Nm) respectively. During SLL, peak
internal rotation knee moments were reduced by 10% (3.2 Nm), while peak flexion knee
moments increased by 56% (121.6 Nm).
Figure 5.4
The critical kinematic changes corresponding to the reduction in valgus knee torque during
the WA phase of UnSS were trunk rotation towards the desired direction of travel (2.9°),
right shoulder adduction (15.7°), left shoulder flexion (4.1°) and right hip abduction (3.1°)
(stance limb, right leg). The critical kinematic changes corresponding to the reduction in
0
50
100
150
200
250
300
350
Flexion Valgus Int. Rotation
Pea
k K
nee
Mom
ent
(Nm
)
Pre
Post
0
50
100
150
200
250
300
350
Flexion Valgus Int. Rotation
Pea
k K
nee
Mo
men
t (N
m)
Pre
Post
Sidestepping SLL
Peak flexion, valgus and internal rotation knee moments pre-to-post kinematic optimization calculated during the WA phase of UnSS (Left) and SLL (Right).
93
valgus knee torque during the WA phase of SLL were left hip (7.8º) and knee (19.3º)
extension (stance limb, right leg).
5.4 DISCUSSION
Results showed the outer-level optimisation method was effective in creating a
dynamically consistent simulation similar to the experimentally recorded motion of the
stance phase of straight-line over-ground running. Results also showed that the RRA can
be used to identify causal relationships between an individual’s whole-body kinematics
and valgus knee loading during the WA phase of both UnSS and SLL.
The outer-level optimisation method was effective in simplifying the RRA process to
produce a dynamically consistent simulation (residuals < 1N and 1 Nm) of human
movement. This method represents a simplified process, as the number of user defined
input parameters was reduced by 96% (74 to 3) for a full-body musculoskeletal model
(DoF = 37). This method also reduced the total time required to generate a simulation
from three 3 research days when using a user’s intuition to 11.8 hours when using the
outer-level optimisation method.
We did not expect a maximum kinematic difference of 3.1º between the outer-level
optimisation and user intuition method. However, we felt that such changes were
necessary for the dramatic reductions in residual forces/torques (151N and 23Nm) when
using the outer-level optimisation method. The magnitude of these differences were
determined by the weighting used with the outer-level optimisation cost function (i.e.
, and ). Given lower weighting values associated with the pelvic residuals
( ) and/or higher weightings for the tracking errors ( and ), the outer-level cost
function would be minimised differently, with emphasis placed on the tracking errors rather
than the pelvic residuals. There is currently no available literature defining an acceptable
error limit between an experimentally recorded and simulation of human motion. With a
maximal kinematic difference less than 4.0º within the 37 DoF model, and the overall
simulated motion being representative of the experimentally recorded motion, we felt this
was an acceptable error limit.
Results showed that in principle, the RRA can be used to optimise an athlete’s whole-body
kinematics to reduce valgus knee loading during UnSS and SLL. The kinematic
94
differences used to reduce valgus knee torque during UnSS were trunk rotation, right
shoulder adduction, left shoulder flexion and right hip abduction. The kinematics used to
reduce valgus knee torque during SLL were to elevated left hip and knee extension.
These results support previous literature, which has shown that trunk posture (Dempsey et
al., 2007), arm position (Chaudhari et al., 2005) and hip neuromuscular control (Kipp et al.,
2011; McLean et al., 2005) are associated with peak frontal, sagittal and/or transverse
knee loading during UnSS and/or SLL. Results also show that there is indeed a
relationship between both upper and lower body kinematic and peak valgus knee torques
during UnSS and SLL, supporting a rationale to look above the hip in development of
lower body prophylactic training protocols (Lloyd, 2006). Adding to the previous literature,
results show that when treating UnSS and SLL as dynamic, multi-body systems, kinematic
changes were never observed in just one DoF, but were was always coupled with
kinematic changes from at least one other joint along the kinematic chain. Future
research with larger sample sizes are needed to determine if these findings are simulation
specific or associated with a consistent and/or a generalised kinematic strategy. It is only
then the efficacy of in-silico technique training to reduce valgus knee loading and ACL
injury in community level athletes can be established.
These methods possess enormous potential within the fields of injury prevention, OA
development and even orthopaedics. For example, with the ability to identify causal
relationships between an athlete’s kinematics and joint loading, more effective training
protocols may be developed to reduce ACL injury risk in community level athletes.
Expanding from methods presented by Fregly et al., (2007), an open-source subject-
specific in-silico technique training method is available to re-train knee OA populations to
walk with in a manner that reduces varus knee moments and risk of disease progression.
Additionally the ability for these in-silico technique training methods to be used as a
treatment for patients following a total knee arthroplasty by teaching them to walk, post-
surgery, with reduced joint loading, we possess the ability to extend the life span of their
knee replacement. It is acknowledged that these methods represent the first step of many
before these applications are observed in society. However, it should be recognized that
with future research, the theoretical applications of in-silico technique training are both
broad and immense.
5.5 CONCLUSIONS
95
The outer-level optimisation technique with the RRA is an appropriate method for creating
dynamically consistent simulations of human motion with minimal complexity. RRA in
OpenSim is an open-source method capable of identifying causal relationships between
an individual’s kinematics and valgus knee loading during both UnSS and SLL.
Reference list chapter 5
Besier, T.F., Sturnieks, D.L., Alderson, J.A., Lloyd, D.G., 2003. Repeatability of gait data using a functional hip joint centre and a mean helical knee axis. J Biomech. 36 (8), 1159-1168. Bisseling, R.W., Hof, A.L., 2006. Handling of impact forces in inverse dynamics. J Biomech. 39 (13), 2438-2444. Caraffa, A., Cerulli, G., Projetti, M., Aisa, G., & Rizzo, A., 1996. Prevention of anterior cruciate ligament injuries in soccer. A prospective controlled study of proprioceptive training. Knee Surg Sports Traumatol Arthrosc. 4 (1), 19-21. Chaudhari, A.M., Hearn, B.K., Andriacchi, T.P., 2005. Sport-dependent variations in arm position during single-limb landing influence knee loading: Implications for anterior cruciate ligament injury. Am J Sports Med. 33 (6), 824-830.
Cochrane, J.L., Lloyd, D.G., Buttfield, A., Seward, H., McGivern, J., 2007. Characteristics of anterior cruciate ligament injuries in Australian football. J Sci Med Sport. 10 (2), 96-104. Delp, S.L., Anderson, F.C., Arnold, A.S., Loan, P., Habib, A., John, C.T., Guendelman, E., Thelen, D.G., 2007. Opensim: Open-source software to create and analyse dynamic simulations of movement. IEEE Trans Biomed Eng. 54 (11), 1940-1950. Delp, S.L., Loan, J.P., Hoy, M.G., Zajac, F.E., Topp, E.L., Rosen, J.M., 1990. An interactive graphics-based model of the lower extremity to study orthopaedic surgical procedures. IEEE Trans Biomed Eng. 37 (8), 757-767. Dempsey, A.R., Lloyd, D.G., Elliott, B.C., Steele, J.R., Munro, B.J., Russo, K.A., 2007. The effect of technique change on knee loads during sidestep cutting. Med Sci Sports Exerc. 39 (10), 1765-1773. Dempsey A.R., Lloyd D.G., Elliott B.C., Steele J.R., Munro B.J., 2012. Whole body kinematics and knee moments that occur during an overhead catch and landing task in sport. Clin Biomech, (In Press and Published Online, dx.doi.org/10.1016/j.clinbiomech.2011.12.001). Donnelly, C.J., Elliott, B.C., Ackland T.R., Doyle T.L.A, Besier T.F., Finch, C.F., Dempsey, A., Lloyd, D.G., 2012. An anterior cruciate ligament injury prevention framework: Incorporating the recent evidence. J Res Sports Med. [In Press, Accepted January, 2012]. Finch, C.F., 2006. A new framework for research leading to sports injury prevention. J Sci Med Sport, 9 (1-2), 3-9.
96
Fregly, B.J., Reinbolt, J.A., Rooney, K.L., Mitchell, K.H., Chmielewski, T.L., 2007. Design of patient-specific gait modifications for knee osteoarthritis rehabilitation. IEEE Trans Biomed Eng. 54 (9), 1687-1695. Gianotti, S. M., Marshall, S. W., Hume, P. A., & Bunt, L., 2009. Incidence of anterior cruciate ligament injury and other knee ligament injuries: a national population-based study. J Sci Med Sport. 12 (6), 622-627. Hamner, S.R., Seth, A., Delp, S.L., 2010. Muscle contributions to propulsion and support during running. J Biomech. 43 (14), 2709-2716. Janssen, K.W., Orchard, J.W., Driscoll, T.R., van Mechelen, W., 2011. High incidence and costs for anterior cruciate ligament reconstructions performed in australia from 2003-2004 to 2007-2008: Time for an anterior cruciate ligament register by scandinavian model? Scand J Med Sci Sports. doi: 10.1111/j.1600-0838.2010.01253.x Kipp, K., McLean, S. G., & Palmieri-Smith, R. M., 2011. Patterns of hip flexion motion predict frontal and transverse plane knee torques during a single-leg land-and-cut maneuver. Clin Biomech (Bristol, Avon). 26 (5), 504-508. Koga, H., Nakamae, A., Shima, Y., Iwasa, J., Myklebust, G., Engebretsen, L., . . . Krosshaug, T., 2010. Mechanisms for noncontact anterior cruciate ligament injuries: knee joint kinematics in 10 injury situations from female team handball and basketball. Am J Sports Med. 38 (11), 2218-2225. Lloyd, D., 2006. Moving away from traditional foci may help us understand sporting performance and injuries. J Sci Med Sport. 9, 275 -276. Markolf, K.L., Burchfield, D.M., Shapiro, M.M., Shepard, M.F., Finerman, G.A., Slauterbeck, J.L., 1995. Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res. 13 (6), 930-935. McLean, S.G., Huang, X., Su, A., Van Den Bogert, A.J., 2004. Sagittal plane biomechanics cannot injure the acl during sidestep cutting. Clin Biomech (Bristol, Avon). 19 (8), 828-838. McLean, S. G., Huang, X., & van den Bogert, A. J., 2005. Association between lower extremity posture at contact and peak knee valgus moment during sidestepping: implications for ACL injury. Clin Biomech (Bristol, Avon). 20 (8), 863-870. McLean, S.G., Huang, X., van den Bogert, A.J., 2008. Investigating isolated neuromuscular control contributions to non-contact anterior cruciate ligament injury risk via computer simulation methods. Clin Biomech (Bristol, Avon). 23 (7), 926-936. Oiestad, B.E., Engebretsen, L., Storheim, K., Risberg, M.A., 2009. Knee osteoarthritis after anterior cruciate ligament injury: A systematic review. Am J Sports Med. 37 (7), 1434-1443. Orchard, J., & Seward, H. (2009). 17th Annual AFL injury Report: 2008. 2010, 1-14. Retrieved from http://www.afl.com.au website: http://www.afl.com.au Quatman, C.E., Kiapour, A., Myer, G.D., Ford., K.R., Demetropoulos, C.K., Goel, V.K., Hewett, T.E., 2011. Cartilage pressure distributions provide a footprint to define female anterior cruciate ligament injury mechanisms. Am J Sports Med. 39 (8), 1706-1713.
97
Shin, C.S., Chaudhari, A.M., Andriacchi., 2011. Valgus plus internal roation moments increase anterior cruciate ligament strain more tahn either alone. Med Sci Sports Exerc. 43 (8), 1484-1491 Thelen, D.G., Anderson, F.C., 2006. Using computed muscle control to generate forward dynamic simulations of human walking from experimental data. J Biomech. 39 (6), 1107-1115. Winter, D., 2005. Motor control of human movement, ed. 3. John Wiley & Sons, Inc., Hoboken, New Jersey.
Withrow, T.J., Hutson, L.J., Wojtys, E.M., Ashton-Miller, J.A., 2006. The effect of an impulsive knee valgus moment on in vitro relative ACL strain during a simulated jump landing. Clin Biomech (Bristol, Avon). 21 (9), 977-83.
98
CHAPTER 6
OPTIMIZING WHOLE-BODY KINEMATIC TO MINIMISE VALGUS KNEE LOADING
DURING SIDESTEPPING: IMPLICATIONS FOR ACL INJURY RISK
This paper has been accepted for publication in the Journal of Biomechanics.
Donnelly, C.J., Elliott, B., Lloyd, D.G. and Reinbolt, J.A. (2012). Optimizing Whole body Kinematics to minimize valgus knee loading during sidestepping: Implications for ACL injury risk. J Biomech. 45:1491-1497, doi:10.1016/j.jbiomech.2012.02.010.
The PhD candidate, Cyril J. Donnelly accounted for 70% of the intellectual property
associated with the final manuscript. Collectively, the remaining authors contributed 30%.
Conflict of Interest: There were no financial or personal relationships with other people or
organizations that could have biased the presented work
99
Abstract
The kinematic mechanisms associated with elevated externally applied valgus knee
moments during non-contact sidestepping and subsequent anterior cruciate ligament
(ACL) injury risk are not well understood. To address this issue, the residual reduction
algorithm (RRA) in OpenSim was used to create nine subject-specific, full-body (37
degrees of freedom) torque-driven simulations of athletic males performing unplanned
sidestep (UnSS) sport tasks. The RRA was used again to produce an optimized kinematic
solution with reduced peak valgus knee torques during the weight acceptance phase of
stance. Pre-to-post kinematic optimization, mean peak valgus knee moments were
significantly reduced by 44.2 Nm (p¼0.045). Nine of a possible 37 upper and lower body
kinematic changes in all three planes of motion were consistently used during the RRA to
decrease peak valgus knee moments. The generalized kinematic strategy used by all nine
simulations to reduce peak valgus knee moments and subsequent ACL injury risk during
UnSS was to redirect the whole-body center of mass medially, towards the desired
direction of travel.
Keywords: Injury; Prevention; Knee; Simulation; Optimization; Technique
6.1 INTRODUCTION Anterior cruciate ligament (ACL) injuries in sport are common (Gianotti et al., 2009;
Janssen et al., 2011). New Zealand and Australia spend approximately 17.4 million NZD
(Gianotti et al., 2009) and 75 million AUD (Janssen et al., 2011) on ACL injuries each year.
Extrapolating from figures reported by Gianotti et al. (2009) and current world population
estimates (World Bank, 2010); the United States annually spend approximately 1 billion
USD on ACL injury management. Approximately 55% of ACL injured athletes are not
capable of returning to the same level of competition two years post reconstruction (Dunn
and Spindler, 2010), a percent that increases to 70% after three years (Roos et al., 1995),
which were over double that of a comparable group of non-ACL injured athletes (Ekstand
et al., 1990; Roos et al., 1995). A rupture to the ACL can be considered one of the most
severe knee injuries an athlete can sustain in sport
More than one half of non-contact ACL injuries occur during sidestepping sport
manoeuvres (Cochrane et al., 2007; Koga et al., 2010; Krosshaug et al., 2007).
Biomechanical studies have shown that during the weight acceptance (WA) phase of
100
sidestepping, which is from initial heel contact to the first trough in the vertical ground
reaction force vector (Dempsey et al., 2007), peak valgus knee moments are up to 2-times
larger than those observed during straight-line running (Besier et al., 2001). During
weightbearing (i.e. stance) (Fleming et al., 2001) and when valgus knee moments are
combined with anterior tibial translations, ACL strain is significantly elevated (Markolf et
al., 1995; Withrow et al., 2006). These are similar to the loading patterns needed to
increase ACL strain and/or reach injurious loading thresholds in-silico (McLean et al.,
2004; McLean et al., 2008; Quatman et al., 2011; Shin et al., 2011). Reducing valgus knee
loading during sport tasks like sidestepping is therefore considered an appropriate
countermeasure to reduce ACL injury risk.
Hewett et al. (2005) has shown peak valgus knee moments during landing are good
predictors of ACL injury. Peak valgus knee moments (Besier et al., 2001; Chaudhari et al.,
2005; Dempsey et al., 2007; McLean et al., 2005) and peak in-vivo ACL strain (Cerulli et
al., 2003) are generally observed during WA. Consequently, one focus of ACL injury
prevention training interventions is to reduce valgus knee moments during the WA phase
of sidestepping (Cochrane et al., 2010; Dempsey et al., 2009), when ACL injury risk is
thought to be the greatest.
Both neuromuscular (Myer et al., 2005) and balance (Cochrane et al., 2010) training have
been shown to reduce valgus knee moments during landing and sidestepping. However,
these studies have not measured and/or identified the kinematic mechanisms contributing
to these observed reductions in knee loading. Hip (McLean et al., 2005), trunk (Dempsey
et al., 2007) and arm kinematics (Chaudhari et al., 2005) have been shown to be
associated with peak valgus knee moments during sidestepping, while lateral trunk
stability has been shown to be associated with rate of ACL injury (Zazulak et al., 2007).
Although associations between upper body biomechanics and knee loading have been
identified, they are heuristic in nature, providing limited causal information when applied to
complex, multi-body, dynamic movements like sidestepping.
Full-body in-silico simulations, with optimisation computational methods have been used
previously to identify causal relationships between whole-body (WB) kinematics and peak
varus knee moments during walking (Fregly et al., 2007). The open-source
musculoskeletal modelling software OpenSim (simtk.org, Stanford, CA) allows for in-silico
simulations of human movement to be created from three-dimensional (3D) motion data.
101
The Residual Reduction Algorithm (RRA) within OpenSim is an optimisation tool capable
of altering a simulation’s kinematics to reduce peak knee joint loading during sidestepping.
Using this modelling framework and these computational tools, our aim was to identify
causal relationships between WB kinematics and peak valgus knee moments during the
WA phase of sidestepping.
6.2 METHODS
The experimental procedure consisted of three phases: 1) experimental motion data
collection; 2) skeletal modelling and residual force/moment reduction; and 3) minimising
peak valgus knee torques by optimising WB kinematics (Figure 6.1).
Figure 6.1
Thirty-four male Western Australian Amateur Football players completed the UWA
sidestepping protocol at 5 ms-1 (Besier et al., 2001; Dempsey et al., 2007). All
experimental procedures were approved by the University of Western Australia Human
Research Ethics Committee and all participants provided their informed written consent
prior to data collection. WB kinematics and ground reaction forces (GRF) were recorded
from a series of straight-line runs, together with pre-planned and unplanned (UnSS)
sidestep trials, as described in Dempsey et al. (2007). Inverse dynamics (ID) was used to
calculate peak valgus knee moments during the WA phase of sidestepping. From this
cohort, nine participants with the largest mean peak valgus knee moments, which always
occurred during UnSS, were chosen for further analysis. The nine participants were 22.0
± 4.3 years of age, with a mean height and body mass of 1.83 ± 0.04 m and 80.8 ± 6.66
kg, respectively.
A
1. VICON to
OpenSim.m
2. Scaling & IK
3. Residual
Reduction
Algorithm 1
1. ↓ valgus
knee torque
2. Residual
Reduction
Algorithm 2
CB
Overview of the experimental procedure: motion data collection (A), skeletal modelling and residual reduction (B) and optimization WB kinematics to minimised peak valgus knee moments (C).
102
A 12-camera 250 Hz VICON MX motion capture system (VICON Peak, Oxford Metrics
Ltd., UK) recorded 3D full-body kinematics (Dempsey et al., 2007). GRF were
synchronously recorded at 2,000 Hz from a single 1.2x1.2-m force platform (Advanced
Mechanical Technology Inc., Watertown, MA.). Kinematic and GRF data were both low
pass filtered at 15 Hz using a zero-lag 4th order Butterworth digital filter in Workstation
(Vicon Peak, Oxford Metrics Ltd., UK). The cut-off frequency was selected based on a
residual analysis (Winter, 2005) and visual inspection. Applying the same filter and cut-off
frequency to the motion and GRF data has been shown to reduce knee joint kinetic
artefacts (Bisseling and Hof, 2006).
Custom biomechanical models in Matlab (Matlab 7.8, The Math Works, Inc., Natick,
Massachusetts, USA), Vicon Bodybuilder (Dempsey et al., 2007) and functional knee and
hip joint methods (Besier et al., 2003)(Appendix B) were used to calculate subject-specific
joint centres. Joint centres, marker trajectories (Appendix E) and GRF (Appendix F) data
were then exported into OpenSim 1.9.1.
A 14 segment, 37 degree-of-freedom (DoF) rigid-linked skeletal models driven by 37 ideal
torque actuators formed the foundation of each simulation (Appendix D). For clarity, joint
torques, not muscles were used to drive each simulation. Twenty-nine of the model’s DoF
have been described previously (Hamner et al., 2010), to which we added 2 DoF wrist
joints (flexion/extension and ulnar/radial deviation) and 3 DoF knee joints
(flexion/extension, internal/external rotation, and varus/valgus). Internal/external rotation
and the varus/valgus DoF of the knee were modelled as universal joints, with the same
centre of rotation, and moved with the flexion/extension DoF, which was modelled as a
planar joint, allowing the tibia to translate relative to the femur as a function of knee flexion
angle (Delp et al., 1990) (Figure 6.2). Segment lengths were scaled to each participant’s
subject-specific joint centre positions, where segment masses and inertial properties were
scaled to each participant’s total body mass in OpenSim.
Inverse kinematics (IK) (Delp et al., 2007) is a global optimisation method (weighted least-
squares) used in OpenSim to calculate a model’s generalised coordinates (i.e. joint
angles) during the WA phase of UnSS. This is done by minimising the squared distances
between the rigid segment markers of the 37 DoF rigid-linked skeletal model and the
experimentally recorded kinematics by adjusting the model’s generalised coordinates.
103
Following IK, the generalised coordinates and experimental GRF measures were used in a
two-step RRA process within OpenSim.
Inconsistencies between a model’s dynamics and experimental GRF measures (∑Fmodel ≠
GRF) called residual forces and moments are often overlooked when using ID (Delp et al.,
2007). Residual forces and moments, present in all biomechanical models, represent
errors and assumptions in the modelling process (i.e. joint centre and inertial estimates).
OpenSim addresses this issue by creating a 6 DoF joint between the pelvis and ground,
holding residual forces and moments not solved for during ID, satisfying Newton’s second
law (∑Fmodel + ∑Fresiduals = GRF). The goal of the RRA is to produce a set of actuator forces
(i.e. joint torques) to generate joint motions that track a desired set of generalised
coordinates, while minimising the model’s residual forces and moments (Delp et al., 2007;
Thelen and Anderson, 2006). The result is simulation that tracks the experimentally
recorded GRF with dynamic consistency.
Figure 6.2
Depiction of 37 DoF, 14 segment full-body rigid-linked skeletal model. The pelvis segment with respect to ground was defined using 3 translations and 3 rotations (6 DoF). A ball-and-socket was used to represent the hip, shoulder and pelvis to trunk/head joints (3 DoF). The wrists were modeled as universal joints (2 DoF). The radial-ulnar, elbow and ankle joints were modeled as revolutes (1 DoF). The knee joint (3 DoF) was modelled as a planar joint in the flexion/extension axis which allowed the tibia to translate as a function of knee flexion angle (Delp et al., 1990); internal/external rotation and abd/adduction were modeled as universal joints.
104
The first step of RRA optimises trunk centre of mass (CoM) position to reduce mean
residual force and moment offsets. The second step reduces these residuals further by
slightly adjusting the model’s generalised coordinates by minimising the sum of three
components ( ): 1) the weighted ( ) squared errors between the experimental (
)
and simulated ( ) accelerations for each DoF of the skeletal model; 2) the
squared residual forces/torques ( ) (n = 6) proportional to their driving excitation ( )
normalized by their maximum residual forces/torques ( ); and 3) the squared joint
torques ( ) for each DoF ( ) in the model, proportional to their driving excitation ( )
normalized by their maximum torques ( ).
(1)
The RRA establishes a set of excitations ( ), driving the six residual forces/torques
( ) and joint torques ( in a simulation. Using numerical integration (i.e.
), the model’s dynamics are generated at each time step. A time varying set of
generalised joint coordinates ( ), residual forces/torques ( ), and joint torques ( are
output, producing a dynamically consistent torque-driven simulation of UnSS. During
RRA, the orientation of the GRF vector is held relative to the stance foot’s CoM. Thus the
location of the centre of pressure is always in the same relative orientation to the stance
foot’s CoM during the RRA process.
The results of RRA depended on what values are chosen for the 74
input parameters, which include maximal joint torques, maximal pelvic residual
forces/torques and kinematic weightings. To produce the best possible dynamically
consistent simulation, input parameters were solved using an outer optimisation method,
which minimised joint torques (i.e. , residual error (i.e.
) and total
kinematic error (Reinbolt et al., 2011). Additional weightings
were placed on the residuals and kinematic errors, meaning the primary goal of the
external optimisation method was to minimise residual and kinematic error during RRA.
Using these methods, peak residual forces and moments were less than 2.5 N and 0.5
105
Nm, respectively. Maximum root mean squared joint coordinate errors were between 1.0 -
8.4º, with a mean of 3.5 ± 2.8 º for all nine simulations. Furthermore, in a randomly
selected subset of subjects, we compared external knee moment traces produced from
RRA with those calculated using ID. Peak knee moments were within ± 5%, and occurred
within ± 0.016 seconds of each other. Given these results we were confident the
simulated UnSS knee moments were consistent with those reported previously in the
literature (Besier et al., 2001; Dempsey et al., 2007; 2009).
The final stage of this procedure was to minimise peak valgus knee moments during the
WA phase of UnSS. This was accomplished by reducing the maximum joint torque ( )
value associated with the knee’s V/V DoF and RRA re-run using the same maximal
torques, maximal residual forces/torques and kinematic weightings solved for using the
external optimisation method along with the experimental GRF measures. For a kinematic
optimisation solution to be deemed acceptable, stance foot translations were limited to 30
mm (Fregly et al., 2007) in all three directions of motion i.e. medial/lateral (M/L),
anterior/posterior (A/P) and inferior/superior (I/S). Kinematic constraints were placed on
the stance foot to restrict foot translations during the RRA process.
(2)
Pre-to-post kinematic optimisation, selected kinematic and kinetic variables were analysed
during WA. Peak valgus, flexion and internal rotation knee torques calculated during RRA
were expressed as externally applied knee moments (Lloyd, 2001). The mean difference
in WB CoM relative to stance foot CoM as well as relative stance foot CoM orientation pre-
to-post kinematic optimisation were calculated in the M/L, A/P and I/S directions.
Independent one-way ANOVAs were used to compare peak mean valgus, flexion and
internal rotation knee moments pre-to-post kinematic optimisation (α = 0.05) (Appendix H).
Independent one-way ANOVAs with a Bonferroni post hoc test were used to determine if
significant differences in WB CoM relative to stance foot CoM were observed pre-to-post
106
kinematic optimisation between the M/L, A/P and I/S directions (α = 0.05). The mean M/L,
A/P and I/S relative error (%) of the stance foot’s CoM trajectory pre-to-post kinematic
optimisation were also calculated.
The mean angular differences for all 37 DoF pre-to-post kinematic optimisation were
calculated in 20% intervals over WA, and kinematic maps created for all nine simulations
(Figure 6.3) (Appendix I). Each map represented the absolute change in joint angles pre-
to-post kinematic optimisation, for each DoF within the skeletal model (n = 37), in each of
the five time intervals within WA. The mean difference of all joints across all time points
during WA was then calculated. Any joint with an angular kinematic change greater than
2σ above the mean was defined as a critical joint coordinate and identified as a kinematic
change that most influenced the observed changes in peak valgus knee moments pre-to-
post kinematic optimisation.
Figure 6.3
Kinematic mapping of a typical simulation representing the absolute kinematic changes (q) from pre-to-post kinematic optimization for all DoF within the skeletal model (N = 37) at 20% intervals during WA of UnSS.
107
6.3 RESULTS
Pre-to-post kinematic optimisation, peak mean valgus knee moments during UnSS were
significantly reduced by 44.2 Nm (106.1 ± 48.6 to 61.9 ± 36.4 Nm) (p = 0.045). Peak
mean flexion and internal rotation knee moments increased by 24.1 Nm (252.2 ± 80.2 to
276.3 ± 69.4 Nm) and 1.1 Nm (7.6 ± 6.9 to 8.7 ± 7.7 Nm) respectively (Figure 6.4).
Figure 6.4
Pre-to-post kinematic optimisation, unique 3D kinematic changes were used by each
simulation to reduce peak valgus knee moments. However, only nine of a possible 37
critical joint coordinates were used by all nine simulations to reduce peak valgus knee
moments during UnSS (Table 6.1) (Appendix J). Two primary kinematic strategies were
used by the simulations to reduce peak valgus knee moments: The first, used by six of the
nine simulations elevated mean ankle plantar flexion by 7.9 ± 5.2º, while the second, used
by all nine simulations, was to reposition WB CoM medially and anteriorly relative to the
stance foot CoM, which was towards desired direction of travel during the UnSS.
Peak mean knee flexion, valgus and internal rotation moments pre-to-post kinematic optimization calculated during the WA phase of an UnSS. Symbol * indicates a
significant change over time (α = 0.05).
108
The mean change in WB CoM relative to stance foot CoM was 3.1 ± 0.8 cm medially, 1.4 ±
1.4 cm anteriorly and 0.2 ± 0.3 cm superiorly. The mean change in WB CoM was
significantly different between the M/L, A/P and I/S directions (p < 0.001). Post hoc
analysis showed that mean changes in WB CoM were significantly greater in the medial
direction relative to anterior (p = 0.003) and superior direction (p < 0.001), while anterior
changes were significantly greater than changes in the superior direction (p = 0.045)
(Figure 6.5).
Figure 6.5
Mean changes in stance foot CoM were limited to -3.7 ± 2.0, 10.5 ± 2.97 and -3.1 ± 2.1
mm in the M/L, A/P and I/S direction respectively. The mean relative error of the stance
foot’s CoM trajectory was 4.0, 23.9 and 5.1% in the M/L, A/P and I/S directions
respectively (Figure 6.6).
Mean peak changes in WB CoM relative to stance foot CoM position pre-to-post kinematic optimization. Anterior and medial changes are towards the desired change of direction pathway. Symbols * and ** indicated a significant change of p < 0.05 and p < 0.01 respectively.
109
Figure 6.6
Mean change in stance foot CoM position (mm) and relative error (%) with respect to the original foot trajectory pre-to-post kinematic optimization. Anterior, medial and superior changes are positive.
110
Sagittal Plane (deg) Frontal Plane (deg) Transverse Plane (deg) WB CoM v Foot CoM (cm) R_Plant_Flex L_Knee_Ext L_Hip_Flex L_Shold_Ext R_ Shold _Ad L_Hip_Ab R_Hip_Ab Trunk_Rot_Med R_ Shold _Int_Rot Ant/Post Med/Lat Sup/Inf
Sim 1 -- -- -- 4.1 15.7 -- 3.1 2.9 -- 1.2 4.8 -0.1
Sim 2 9.4 7.1 6.8 -- 16.1 -- -- -- -- 4.6 3.0 0.1
Sim 3 3.8 -- -- -- 9.4 -- -- -- -- 0.9 2.3 0.3
Sim 4 5.4 -- -- -- 24.9 -- -- -- -- 0.8 2.5 0.1
Sim 5 -- -- 1.7 -- -- 2.0 -- 3.2 1.0 2.4 3.6 0.2
Sim 6 16.5 -- -- -- 16.2 -- -- -- -- 2.0 2.5 1.0
Sim 7 -- 6.1 -- 8.3 6.2 -- -- -- -- -0.5 2.9 0.0
Sim 8 10.0 2.6 -- -- -- 6.6 -- -- -- 1.0 3.7 0.3
Sim 9 2.2 -- -- -- 5.1 -- -- -- -- 0.4 2.9 0.0
μ 7.9 5.3 4.3 6.2 13.4 4.3 3.1 3.1 1.0 1.4 3.1 0.2
σ 5.2 2.4 3.6 3.0 6.9 3.3 -- 0.2 -- 1.4 0.8 0.3
n 6 3 2 2 7 2 1 2 1 9 9 9
Table 6.1
Individual simulation (Sim), mean (μ) differences of critical joint coordinates (deg) and mean WB CoM position relative to stance foot CoM position (m) pre-to-post kinematic optimization. Anterior, medial and superior changes in degrees are positive. Anterior and medial are both towards the desired change of direction pathway. The symbol "--" means the variable was not identified as a critical joint coordinate.
111
6.4 DISCUSSION
Associations between upper body posture and peak valgus knee moments during
sidestepping have been reported previously in the literature (Chaudhari et al., 2005;
Dempsey et al., 2007; McLean et al., 2005). For example, lateral trunk flexion (Dempsey
et al., 2007) and constraining an athlete’s arms to their mid-line (Chaudhari et al., 2005)
likely restricted their upper body CoM from moving medially during sidestepping, resulting
in the observed increases in peak valgus knee moments. Results from this study confirm
that upper body kinematics indeed influence valgus knee loading during sidestepping.
However, unlike previous findings, results showed that one kinematic change was always
coupled with kinematic changes from at least one other joint along the kinematic chain.
Additionally, results showed that both upper and lower body kinematic changes in all three
planes of motion can be utilized to decrease peak valgus knee loading during UnSS. The
generalised kinematic strategy used by all nine simulations to reduce peak valgus knee
moments during UnSS was to reposition WB CoM medially, towards the desired direction
of travel.
Statistically significant reductions in peak valgus knee moments were accompanied by
increases in both peak flexion and internal rotation knee moments. Increases in flexion
knee moments combined with decreases in peak varus knee moments have been
observed following gait re-training in clinical settings (Fregly et al., 2007; Walter et al.,
2010). In addition, elevated applied flexion moments in isolation are unlikely to reach an
injurious loading threshold in-silico (McLean et al., 2004), while the observed increases in
internal rotation knee moments are considered negligible. Results from this and previous
literature suggest that the in-silico changes in knee moments are consistent with clinical
findings and were effective in reducing surrogate measures of ACL injury risk.
As with all optimisation based research, an enormous solution space exists. As such,
unique kinematic strategies were used by each simulation to reduce peak valgus knee
moments during UnSS. Though the results showed each simulation consistently used the
same nine of a possible 37 joint DoF to reduce peak valgus knee moments during UnSS,
511 (29-1) kinematic combinations remain. The experimental and computational time
required to process a single simulation currently takes approximately 36 hours to
complete, limiting the application of current in-silico subject-specific technique training
methods to “high risk” athletic populations. Future research is therefore needed to
112
develop clinically-relevant ACL injury risk estimates to identify “high-risk” athletes if current
subject-specific in-silico technique training methods can be effectively utilized.
Generalized kinematic strategies to reduce peak valgus knee loading during sidestepping
must be developed for ACL injury risk to be reduced in heterogeneous athletic
populations. A generalised kinematic solution would make it possible for coaches and/or
clinicians to train athletes to sidestep with reduced valgus knee loading. In-silico patient-
specific gait modifications have been successfully used to re-train a high functioning
osteoarthritis (OA) patient to walk with reduced peak adduction (varus) knee moments and
OA related knee pain (Fregly et al., 2007). “Medial-thrust gait”, which in general terms
focuses on increasing support limb flexion and decreasing the size of the moment arm
between the knee joint centre and GRF vector during stance was the generalised
kinematic strategy identified by Fregly et al. (2007). “Medial-thrust gait” training has since
been proven effective in reducing peak varus knee moments in both a single healthy male
(Schache et al., 2008) and elderly male OA patient (Walter et al., 2010).
From the nine critical joint coordinates used by each simulation to reduce valgus knee
moments, two generalised kinematic strategies were identified. One strategy involved
increasing stance foot plantar flexion, while the second was to re-direct the WB CoM
medially, towards the desired change of direction pathway.
The ankle plantar flexion strategy used by six of the nine simulations likely reduced peak
valgus knee moments by changing the position of the ankle joint centre relative to the GRF
vector during WA. Small changes in joint centre position have non-linear effects on
proximal joint torques along the kinematic chain (Reinbolt et al., 2007). Changing joint
centre position also has differing effects on joint torques expressed in the M/L and A/P
DoF (Reinbolt et al., 2007). These non-linear relationships make it difficult to identify how
plantar flexion influences valgus knee moments during sidestepping. Additionally, without
a foot-contact model, it is unlikely that these results would be observed in an experimental
setting. We are therefore limited in our ability to make conclusions associated with
plantar/dorsi flexion and valgus knee loading, leaving this relationship to be verified with
future research.
Re-positioning the CoM medially, towards the desired change of direction pathway is a
motor control strategy used during change-of-direction tasks (Patla et al., 1999) and
113
similar to one of three technique recommendations used to reduce peak valgus knee
moments during sidestepping (Dempsey et al., 2009); meaning this kinematic strategy can
be learned by athletic populations. A secondary benefit of this generalised technique
recommendation is that individuals can develop unique motor control strategies to
successfully learn this technique modification. This generalised technique modification
subsequently represents a form of subject-specific technique training.
One concern for using RRA to reduce peak valgus knee moments during UnSS is that the
sidestep motion may not be preserved pre-to-post kinematic optimisation. These
concerns are addressed in three ways. First, the goal of RRA is to reduce the residual
forces and moments held in the pelvis, producing a torque driven simulation that is
dynamically consistent with the experimental GRF’s measured during UnSS. This is an
important consideration, as these external forces are needed to redirect the WB CoM
during sidestepping (Jindrich et al., 2006) and are therefore a fundamental component of a
realistic simulation of the sidestep motion. Second, the motion of all nine simulation’s
CoM were directed medially, towards the desired direction of travel, making an UnSS look
more like a pre-planned sidestep (Houck et al., 2006). Third, supplementary video data
published with this manuscript shows the sidestep motion was indeed maintained pre-to-
post kinematic optimisation (Appendix J).
Previous literature has shown that in-silico technique modifications are effective in
reducing peak varus knee moments in multiple case studies (Fregly et al., 2007; Schache
et al., 2008; Walter et al., 2010). Findings from this study must now be tested in a
controlled laboratory setting, with large heterogeneous athletic populations. Once the
efficacy of directing the WB CoM medially during sidestepping to reduce peak valgus knee
moments, it can then be recommended to heterogeneous athletic populations.
These methods possess an enormous potential within the injury prevention literature. We
were capable of identifying a single kinematic solution to reduce valgus knee loading
during a complex, multi-body, dynamic movement with an enormous solution space.
Nevertheless, we encourage future in-silico research to build upon these findings. For
example, with a foot contact model, additional kinematic strategies to reduce valgus knee
moments during UnSS and ACL injury risk may be identified. Alternate solutions may also
be possible if the optimisation criterion was amended to both reduce valgus knee loading
and optimise sidestep performance. It is through this rigor that additional casual
114
information may become available and assist in the development of short, concise and
effective technique training protocols designed to reduce ACL injury risk. It is through this
process we can more effectively translate ACL focused research into injury prevention
practice for the community level athlete (Finch, 2006).
Acknowledgments
The authors would like to acknowledge the assistance of Prof. Caroline Finch, Dr Tim
Doyle and Dr Dara Twomey. We thank the Australian National Health and Medical
Research Council (grant number 400937 to Prof. Finch, Prof. Lloyd and Prof Elliott) and
the Western Australian Medical Health and Research Infrastructure Fund (Prof. Lloyd) for
their support of this study. CJ Donnelly would like to thank the Canadian Society for
Biomechanics and The University of Western Australia convocation office for funding his
travel to The University of Tennessee and making this research collaboration possible.
Reference list chapter 6
Besier, T.F., Lloyd, D.G., Cochrane, J.L., Ackland, T.R., 2001. External loading of the knee joint during running and cutting maneuvers. Med Sci Sports Exerc. 33 (7), 1168-1175. Besier, T.F., Sturnieks, D.L., Alderson, J.A., Lloyd, D.G., 2003. Repeatability of gait data using a functional hip joint centre and a mean helical knee axis. J Biomech. 36 (8), 1159-1168. Bisseling, R.W., Hof, A.L., 2006. Handling of impact forces in inverse dynamics. J Biomech. 39 (13), 2438-2444. Cerulli, G., Benoit, D.L., Lamontagne, M., Caraffa, A., Liti, A., 2003. In vivo anterior cruciate ligament strain behaviour during a rapid deceleration movement: Case report. Knee Surg Sports Traumatol Arthrosc. 11 (5), 307-311. Chaudhari, A.M., Hearn, B.K., Andriacchi, T.P., 2005. Sport-dependent variations in arm position during single-limb landing influence knee loading: Implications for anterior cruciate ligament injury. Am J Sports Med. 33 (6), 824-830. Cochrane J.L., Lloyd D.G., Besier T.F., Elliott B.C., Doyle T.L., Ackland T.R., 2010. Training affects knee kinematics and kinetics in cutting maneuvers in sport. Med Sci Sports Exerc. 42 (8), 1535-44. Cochrane, J.L., Lloyd, D.G., Buttfield, A., Seward, H., McGivern, J., 2007. Characteristics of anterior cruciate ligament injuries in australian football. J Sci Med Sport. 10 (2), 96-104. Delp, S.L., Anderson, F.C., Arnold, A.S., Loan, P., Habib, A., John, C.T., Guendelman, E., Thelen, D.G., 2007. Opensim: Open-source software to create and analyse dynamic simulations of movement. IEEE Trans Biomed Eng. 54 (11), 1940-1950.
115
Delp, S.L., Loan, J.P., Hoy, M.G., Zajac, F.E., Topp, E.L., Rosen, J.M., 1990. An interactive graphics-based model of the lower extremity to study orthopaedic surgical procedures. IEEE Trans Biomed Eng. 37 (8), 757-767. Dempsey, A.R., Lloyd, D.G., Elliott, B.C., Steele, J.R., Munro, B.J., 2009. Changing sidestep cutting technique reduces knee valgus loading. Am J Sports Med. 37 (11), 2194-2200. Dempsey, A.R., Lloyd, D.G., Elliott, B.C., Steele, J.R., Munro, B.J., Russo, K.A., 2007. The effect of technique change on knee loads during sidestep cutting. Med Sci Sports Exerc. 39 (10), 1765-1773. Dunn, W. R. & Spindler, K. P., 2010. Predictors of activity level 2 years after anterior cruciate ligament reconstruction (ACLR): a Multicentre Orthopaedic Outcomes Network (MOON) ACLR cohort study. Am J Sports Med, 38, 2040-50. Ekstrand, J., Roos, H., Tropp, H., 1990. Normal course of events amongst Swedish soccer players: an 8-year follow-up study. Br J Sports Med. 24, 117-119. Finch, C., 2006. A new framework for research leading to sports injury prevention. J Sci Med Sport. 9 (1-2), 3-9; discussion 10. Fleming, B.C., Renstrom, P.A., Beynnon, B.D., Engstrom, B., Peura, G.D., Badger, G.J., Johnson, R.J., 2001. The effect of weightbearing and external loading on anterior cruciate ligament strain. J Biomech. 34 (2), 163-170. Fregly, B.J., Reinbolt, J.A., Rooney, K.L., Mitchell, K.H., Chmielewski, T.L., 2007. Design of patient-specific gait modifications for knee osteoarthritis rehabilitation. IEEE Trans Biomed Eng. 54 (9), 1687-1695. Gianotti, S.M., Marshall, S.W., Hume, P.A., Bunt, L., 2009. Incidence of anterior cruciate ligament injury and other knee ligament injuries: A national population-based study. J Sci Med Sport. 12 (6), 622-627. Hamner, S.R., Seth, A., Delp, S.L., 2010. Muscle contributions to propulsion and support during running. J Biomech. 43 (14), 2709-2716. Houck, J.R., Duncan, A., De Haven, K.E., 2006. Comparison of frontal plane trunk kinematics and hip and knee moments during anticipated and unanticipated walking and side step cutting tasks. Gait Posture. 24 (3), 314-322. Hewett, T.E., Myer, G.D., Ford, K.R., Heidt, R.S.J., Colosimo, A.J., McLean, S.G., van den Bogert, A.J., Paterno, M.V., Succop, P., 2005. Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes. A prospective study. Amer J Sports Med. 33(4), 492-501. Jindrich DL, Besier TF, Lloyd DG., 2006. A hypothesis for the function of braking forces during running turns. J Biomech. 39 (9), 1611-1620. Janssen, K.W., Orchard, J.W., Driscoll, T.R., van Mechelen, W., 2011. High incidence and costs for anterior cruciate ligament reconstructions performed in australia from 2003-2004 to 2007-2008: Time for an anterior cruciate ligament register by scandinavian model? Scand J Med Sci Sports. doi: 10.1111/j.1600-0838.2010.01253.x
116
Koga, H., Nakamae, A., Shima, Y., Iwasa, J., Myklebust, G., Engebretsen, L., Bahr, R., Krosshaug, T., 2010. Mechanisms for noncontact anterior cruciate ligament injuries: Knee joint kinematics in 10 injury situations from female team handball and basketball. Am J Sports Med. 38 (11), 2218-2225. Krosshaug, T., Nakamae, A., Boden, B.P., Engebretsen, L., Smith, G., Slauterbeck, J.R., Hewett, T.E., Bahr, R., 2007. Mechanisms of anterior cruciate ligament injury in basketball: Video analysis of 39 cases. Am J Sports Med. 35 (3), 359-367. Lloyd, D.G., 2001. Rationale for training programs to reduce anterior cruciate ligament injuries in australian football. J Orthop Sports Phys Ther. 31 (11), 645-654; discussion 661. Markolf, K.L., Burchfield, D.M., Shapiro, M.M., Shepard, M.F., Finerman, G.A., Slauterbeck, J.L., 1995. Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res. 13 (6), 930-935. McLean, S.G., Huang, X., Su, A., Van Den Bogert, A.J., 2004. Sagittal plane biomechanics cannot injure the acl during sidestep cutting. Clin Biomech (Bristol, Avon). 19 (8), 828-838. McLean, S.G., Huang, X., van den Bogert, A.J., 2005. Association between lower extremity posture at contact and peak knee valgus moment during sidestepping: Implications for acl injury. Clin Biomech (Bristol, Avon). 20 (8), 863-870. McLean, S.G., Huang, X., van den Bogert, A.J., 2008. Investigating isolated neuromuscular control contributions to non-contact anterior cruciate ligament injury risk via computer simulation methods. Clin Biomech (Bristol, Avon). 23 (7), 926-936. Myer GD, Ford KR, Palumbo JP, Hewett TE., 2005. Neuromuscular training improves performance and lower-extremity biomechanics in female athletes. J Strength Cond Res. 19 (1), 51-60. Patla, A.E., Adkin, A., Ballard, T., 1999. Online steering: Coordination and control of body centre of mass, head and body reorientation. Exp Brain Res. 129 (4), 629-634. Quatman, C.E., Kiapour, A., Myer, G.D., Ford., K.R., Demetropoulos, C.K., Goel, V.K., Hewett, T.E., 2011. Cartilage pressure distributions provide a footprint to define female anterior cruciate ligament injury mechanisms. Am J Sports Med. 39 (8), 1706-1713. Reinbolt, J.A., Haftka, R.T., Chmielewski, T.L., Fregly, B.J., 2007. Are patient-specific joint and inertial parameters necessary for accurate inverse dynamics analyses of gait? IEEE Trans Biomed Eng. 54 (5), 782-793. Reinbolt, J.A., Seth, A. and Delp, S.L. "Simulation of human movement: applications using OpenSim." Procedia IUTAM, 2(1):186–198, 2011. Roos, H., Ornell, M., Gardsell, P., Lohmander, L.S., Lindstrand, A., 1995. Soccer after anterior cruciate ligament injury--an incompatible combination? A national survey of incidence and risk factors and a 7-year follow-up of 310 players. Acta Orthop Scand. 66 (2), 107-112.
117
Schache, A.G., Fregly, B.J., Crossley, K.M., Hinman, R.S., Pandy, M.G., 2008. The effect of gait modification on the external knee adduction moment is reference frame dependent. Clin Biomech (Bristol, Avon). 23 (5), 601-608. Shin, C.S., Chaudhari, A.M., Andriacchi., 2011. Valgus plus internal roation moments increase anterior cruciate ligament strain more tahn either alone. Med Sci Sports Exerc. 43 (8), 1484-1491. Thelen, D.G., Anderson, F.C., 2006. Using computed muscle control to generate forward dynamic simulations of human walking from experimental data. J Biomech. 39 (6), 1107-1115. The World Bank Group [Internet]. Washington, DC (USA): World Population Estimates; [cited 2010 June 7]. Available from: http://data.worldbank.org. Walter, J.P., D'Lima, D.D., Colwell, C.W., Jr., Fregly, B.J., 2010. Decreased knee adduction moment does not guarantee decreased medial contact force during gait. J Orthop Res. 28 (10), 1348-1354. Winter, D., 2005. Motor control of human movement, ed. 3. John Wiley & Sons, Inc., Hoboken, New Jersey. Withrow, T.J., Hutson, L.J., Wojtys, E.M., Ashton-Miller, J.A., 2006. The effect of an impulsive knee valgus moment on in vitro relative ACL strain during a simulated jump landing. Clin Biomech (Bristol, Avon). 21 (9), 977-83. Zazulak, B.T., Hewett, T.E., Reeves, N.P., Goldberg, B., Cholewicki, J., 2007. Deficits in neuromuscular control of the trunk predict knee injury risk: A prospective biomechanical-epidemiologic study. Am J Sports Med. 35 (7), 1123-1130.
118
CHAPTER 7
SUMMARY AND CONCLUSIONS
7.1 THESIS GOALS
The two primary goals of this thesis were to: 1) determine if balance and technique training
when implemented in ‘real-world’ community level training environments was effective in
reducing peak knee loading and/or increasing muscular support during pre-planned and
unplanned sidestepping tasks. 2) Develop an open-source method to identify causal
relationships between an athlete’s kinematics and knee joint loading during sidestepping
and single-leg landing.
Results showed that balance and technique training was not effective in reducing external
knee moments and/or changing the activation of the muscles crossing the knee during
pre-planned and unplanned sidestepping when implemented adjunct to normal ‘real-world’
Australian football training setting. From these findings, it is apparent that much work is
needed before the positive laboratory based biomechanical training effects of plyometric,
balance, resistance and/or technique training in reducing peak knee loading and/or
increasing muscular support (Chappell & Limpisvasti, 2008; Cochrane et al., 2010;
Dempsey et al., 2009; Hewett et al., 1996; Myer et al., 2005; Myer et al., 2006; Lim et al.,
2009; Zebis et al., 2008) are observed in ‘real-world’ training environments. However, we
hope future research utilizes the training and biomechanical testing protocols used in
chapters three and four, with the ACL injury prevention framework proposed in chapter two
to help refine the development and implementation of future community level ACL injury
prevention training programs. It is evident future research focused on understanding
athlete perceptions of, and compliance to, biomechanically based ACL injury prevention
protocols is needed (Finch et al., 2011). Additionally, it is also apparent a coach’s
attitudes and beliefs toward an intended ACL injury prevention need to be addressed
before the intended benefits associated with a prophylactic training intervention can be
effectively translated to athletes in ‘real-world’ community level training environments
(Finch, 2006; Finch et al., 2011; Twomey et al., 2009).
119
An open-source method to identify causal links between an athlete’s kinematics and
valgus knee joint loading during sidestepping and single-leg landing was developed in
chapters five and six. Using these methods, we have shown that re-directing an athlete’s
whole-body centre of mass towards the desired change of direction pathway during
sidestepping is a generalised kinematic strategy that can be used to reduce valgus knee
loading and subsequent ACL injury risk. It is by identifying these types of casual
relationships we will be better able to develop ACL injury prevention training protocols that
target the factors associated with ACL injury risk and more effectively transfer positive
laboratory-based training effects to ‘real-world’ training environments. It is then we may
observe reductions of ACL injury rates across heterogeneous athletic populations in the
future.
7.2 SPECIFIC AIMS AND HYPOTHESES
7.2.1 Chapter 2: An anterior cruciate ligament injury prevention framework:
Incorporating the recent evidence
The aims of this study were to develop an ACL injury prevention framework specific to
the intrinsic factors associated with non-contact ACL injuries and provide a rationale for
the design of ACL injury prevention training protocols. Incorporating the most recent
empirical ACL focused research we have developed an injury prevention framework
specific to the ACL. Within this framework we have provided a rationale for the design
of injury prevention training protocols. In the development of this framework we also
identified gaps in the literature. First, there is little to no research testing the
effectiveness of ACL injury prevention training protocols in reducing peak joint loading
and/or increases muscle support when implemented in ‘real-world’ training
environments. Secondly, there is a lack of research identifying the biomechanical
mechanisms by which training acts to reduce injury risk and/or why a training protocol
was associated with a positive or inconclusive training outcome. By identifying these
causal relationships, more effective injury prevention training programs can be
developed, and in turn reduce ACL injury rates in the future.
120
7.2.2 Chapter 3: Changes in knee joint biomechanics following balance and
technique training and a season of Australian Football
The aims of this study were to determine if balance and technique training,
implemented adjunct to pre-season and regular season Australian football training was
effective in reducing peak knee moments during the weight acceptance phase of pre-
planned and unplanned sidestepping. The secondary aim was to determine if an
Australian football player’s knee joint biomechanics changed over a season of
Australian football. The major finding of this study was that 28 weeks of balance and
technique training implemented in a ‘real-world’ community level training environment
was not effective in changing an athlete’s knee joint biomechanics during either pre-
planned or unplanned sidestepping. However, knee moments during both pre-planned
and unplanned sidestepping tasks were found to respond differently over the playing
season. These results showed that pre-planned and unplanned sidestepping are
unique sporting tasks and should both be used when assessing the effectiveness of
prophylactic training interventions.
Hypotheses
Balance and technique training will reduce both peak valgus and internal rotation
knee moments during the weight acceptance phase of anticipated and
unanticipated sidestepping.
There were no differences in peak valgus and internal rotation knee moments between
the balance and technique training group and ‘sham’ training group (control). Therefore
this hypothesis was rejected.
Peak valgus and internal rotation knee moments during the weight acceptance
phase of anticipated and unanticipated sidestepping will not change over a season
of Australian Football.
Both training groups displayed a significant decrease in peak internal rotation knee
moments during pre-planned sidestepping following a season of Australian football.
Both training groups displayed a significant increase in peak valgus knee moments
during unplanned sidestepping following a season of Australian football. Therefore this
hypothesis was rejected.
121
7.2.3 Chapter 4: Changes in muscle activation following balance and technique
training and a season of Australian Football
The primary aim of this study was to determine if balance and technique training
implemented adjunct to pre-season and regular season Australian football training
influenced the activation patterns of the muscles crossing the knee during pre-planned
and unplanned sidestepping. The secondary purposes were to: 1) determine if an
Australian football player’s muscle activation changed over a season of Australian
football, 2) determine if changes in muscle activation following balance and technique
training were proportional to changes in knee loading during pre-planned and
unplanned sidestepping and 3) determine if changes in muscle activation following a
season of Australian football are proportional to changes in knee loading during pre-
planned and unplanned sidestepping.
The major finding of this study was that 28 weeks of balance and technique training
implemented in a ‘real-world’ community level training environment was not effective in
changing the activation patterns of the muscles crossing the knee during either pre-
planned or unplanned sidestepping. However, results did shown that following a
season of Australian Football, total muscle activation significantly increased, while the
co-contraction ratios were directed towards muscles that could support both applied
flexion and valgus knee moments. When analysing changes in muscle activation and
knee loading together, results showed muscle activation patterns following a season of
Australian football were better suited to protect the knee and ACL from external loading
during pre-planned sidestepping when compared with unplanned sidestepping. These
results suggests an athlete may be at increased risk of ACL injury when conducting
unplanned sports tasks in the latter half of a playing season.
Hypotheses
Balance and technique training will:
i. Increase the total muscle activation of the muscles crossing the knee during
the pre-contact phase of pre-planned and unplanned sidestepping.
122
There were no differences in total muscle activation between the balance and
technique training group and ‘sham’ training group (control). Therefore this hypothesis
was rejected.
ii. Increase the co-contraction between knee flexors and extensors during the
pre-contact phases of pre-planned and unplanned sidestepping.
Flexor/extensor co-contraction ratios were not different between the balance and
technique training group and ‘sham’ training group (control). Therefore this hypothesis
was rejected.
iii. Increase the relative activation of muscles with medial moment arms during
pre-planned sidestepping.
Medial/Lateral co-contraction ratios were not different between the balance and
technique training group and ‘sham’ training group (control). Therefore this hypothesis
was rejected.
The total activation of the muscles crossing the knee during the pre-contact and
weight acceptance phases of pre-planned and unplanned sidestepping will not
change over a season of Australian football.
Total muscle activation was significantly elevated during the pre-contact and weight
acceptance phases of pre-planned and unplanned sidestepping following a season of
Australian football. Therefore this hypothesis was rejected.
The directed co-contraction ratios of the muscles crossing the knee during the pre-
contact and weight acceptance phases of pre-planned and unplanned sidestepping
will not change over a season of Australian football.
Co-contraction ratios were directed towards muscles with extensor moment arms
following a season of Australian football during the pre-contact and weight acceptance
phases of pre-planned and unplanned sidestepping. The relative activation of the
semimembranosus significantly increased following a season of Australian football
123
during the pre-contact and weight acceptance phases of pre-planned and unplanned
sidestepping. Therefore this hypothesis was rejected.
Pre-contact total muscle activation following balance and technique training will be
greater than changes in knee loading during the weight acceptance phase of pre-
planned and unplanned sidestepping.
There were no differences in total muscle activation or co-contraction ratios between
the balance and technique training group and ‘sham’ training group (control).
Therefore this hypothesis was rejected.
Pre-contact total muscle activation following a season of Australian football will be
similar to changes in knee loading during the weight acceptance phase of pre-
planned and unplanned sidestepping.
Pre-contact total muscle activation significantly increased, while peak flexion knee
moments remained unchanged during the weight acceptance phase of pre-planned
and unplanned sidestepping. This hypothesis is therefore confirmed for sagittal plane
knee moments.
Pre-contact total muscle activation was lower during unplanned sidestepping when
compared with pre-planned sidestepping. Peak valgus knee moments were
significantly higher during the weight acceptance phase of unplanned sidestepping
when compared with pre-planned sidestepping. This hypothesis is therefore rejected
for frontal plane knee moments during unplanned sidestepping and confirmed for pre-
planned sidestepping.
7.2.4 Chapter 5: An open-source computational method to optimise simulated
human motion to reduce valgus knee loading during sidestepping and single-leg
landing.
The aims of this study were to develop a simplified computational method to create
dynamically consistent simulations of human motion. Using this method the second
aim was to develop a method utilizing the Residual Reduction Algorithm in OpenSim to
optimise a simulation’s kinematics to minimise peak valgus knee torques during the
124
weight acceptance phase of sidestepping and single-leg landing. An outer-level
optimisation with the Residual Reduction Algorithm was effective in creating a
dynamically consistent (residuals < 1N and 1 Nm) simulation of the stance phase of
straight-line over-ground running similar to the experimentally recorded kinematics
(peak RMS kinematic error < 4°). The Residual Reduction Algorithm was capable of
identifying causal relationships between a simulation’s kinematics and peak valgus
torque during the weight acceptance phase of both sidestepping and single-leg landing.
These methods have provided the literature with an open-source simplified method to
create dynamic simulations of human movement and identify causal relationships
between a simulation’s kinematics and peak joint loading during dynamic sporting
tasks.
Hypotheses
The Residual Reduction Algorithm in OpenSim with an outer-level optimisation
method can be used to create dynamically consistent simulations of human motion.
Results showed the outer-level optimisation method was effective in creating a
dynamically consistent simulation (residuals < 1N and 1 Nm) similar to the
experimentally recorded motion of the stance phase of straight-line over-ground
running (RMS kinematic error < 4°). This hypothesis is therefore confirmed.
The Residual Reduction Algorithm in OpenSim can be used to identify causal links
between a simulation’s whole-body kinematics and valgus knee moments during
the weight acceptance phase of sidestepping.
The Residual Reduction Algorithm in OpenSim can be used to optimise a simulation’s
whole-body kinematics to minimise valgus knee loading during sidestepping. This
hypothesis is therefore confirmed.
The Residual Reduction Algorithm in OpenSim can be used to identify causal links
between a simulation’s whole-body kinematics and valgus knee moments during
the weight acceptance phase of single-leg landing.
125
The Residual Reduction Algorithm in OpenSim can be used to optimise a simulation’s
whole-body kinematics to minimise valgus knee loading during single-leg landing. This
hypothesis is therefore confirmed.
7.2.5 Chapter 6: Optimizing whole-body kinematics to minimise valgus knee loading
during sidestepping: implications for ACL injury risk.
The aim of this study was to use an outer-level optimisation technique, the open-source
musculoskeletal modelling platform OpenSim and the Residual Reduction Algorithm to
identify a generalised kinematic strategy to reduce peak valgus knee moments during
the weight acceptance phase of unplanned sidestepping. The generalised kinematic
strategy identified to reduce peak valgus knee moments during the weight acceptance
phase of unplanned sidestepping was to reposition the whole body centre of mass
medially, towards the desired direction of travel. This generalised technique
modification represents a form of subject-specific technique training and is capable of
being implemented in community level training environments.
Hypotheses
Frontal plane upper body kinematics will be related to increased peak valgus knee
moments during unplanned sidestepping.
The generalised kinematic strategy used by all simulations to minimised valgus knee
moments during the weight acceptance phase of unplanned sidestepping was to
reposition the whole body centre of mass medially. This hypothesis is therefore
confirmed.
Multiple kinematic changes along the kinematic chain will be used to minimising
peak valgus knee moments during the weight acceptance phase of unplanned
sidestepping.
Each simulation used at a minimum of two kinematic changes to reduce valgus knee
moments during the weight acceptance phase of unplanned sidestepping. This hypothesis
is therefore confirmed.
126
7.3 SUMMARY OF STUDY LIMITATIONS
During the balance and technique training intervention, we did not measure
athlete’s perceptions of, and compliance to the training protocol. We also did not
measure a coach’s attitudes and beliefs toward the injury prevention program.
With these measurements available, we may have been able to better identify the
factors contributing to the non-significant findings presented in chapters 3 and 4.
We did not model the anterior cruciate ligament (ACL). From the cadaveric
literature, we used externally applied flexion, valgus knee and internal rotation
knee moments as surrogate measures of ACL strain and injury risk. With more
sophisticated modelling techniques, future research may be able to develop
subject-specific models capable of quantifying the complex interaction between
joint contact, muscle force estimates, knee joint loading and ultimately ACL strain.
We used sEMG and moment arm estimates from the cadaveric literature to
estimate muscle function during sporting tasks. Again, due to limitations in our
ability to estimate muscle force during high velocity, dynamic sporting tasks, future
research is needed to bridge this gap. With this information we may be able to
provide more informed or appropriate neuromuscular muscle activation strategies
to support the knee and reduce ACL injury risk.
Findings presented in chapter 5 and 6 were made following an in-silico analysis.
Findings from this study must now be tested in a controlled laboratory setting, with
large heterogeneous athletic populations. Once the efficacy of these in-silico
findings are verified experimentally, they can then be recommended to larger scale
heterogeneous community level athletic populations.
Findings from chapters 5 and 6 were consistent with reducing valgus knee loading
during sidestepping. Reducing anterior drawer forces and/or internal rotation
moments were not tested. Alternate solutions may be available if the optimisation
criterion was amended to reduce anterior drawer forces, valgus and internal
rotation moments simultaneously.
We did not use a foot contact model during our simulation process in chapters 5
and 6.
127
7.4 FUTURE RESEARCH
Currently the majority of literature within the ACL injury prevention framework (Chapter 2,
Figure 2.4) proposed in chapter two lie within stages one through four. Training
interventions conducted in ‘ideal’ laboratory based environments have been shown to be
effective in reducing peak knee moments and/or increasing muscular support during the
weight acceptance phase of sidestepping and single-leg landing (Chappell & Limpisvasti,
2008; Cochrane et al., 2010; Dempsey et al., 2009; Hewett et al., 1996; Myer et al., 2005;
Myer et al., 2006; Lim et al., 2009; Zebis et al., 2008). When conducted in ‘real-world’
training environments, as done in chapters three and four, 28 weeks of balance and
technique training were found to be ineffective or unsuccessful in reducing peak knee
moments and/or increasing muscular support during pre-planned and unplanned
sidestepping sport tasks. What was evident from these studies is that future research is
needed to evaluate the challenges associate with implementing effective ‘real-world’
training interventions within community level training environments. Specifically, future
research should focus on an athlete perceptions of, and compliance to, biomechanically
based ACL injury prevention protocols (Finch et al., 2011). Of equal importance are a
coach’s attitudes and beliefs toward implementing an ACL injury prevention program.
These factors must be addressed if the intended benefits of a prophylactic training
protocol are effectively translated to the athlete (Twomey et al., 2009). Positive attitudes
with reference to the benefits of a prophylactic training protocol from both athletes and
coaches are also needed for a community level athlete to comply with a given training
protocol (Deci and Ryan, 1985). It is then the positive biomechanical outcomes
associated to a prophylactic training intervention can be transferred to ‘real-world’
community level training environments (Finch, 2006; Finch et al., 2011; Twomey et al.,
2009).
From the ACL injury prevention framework presented in chapter two, it is apparent future
research is needed to develop clinically-relevant screening tools to identify athletes that
may be at increased risk of ACL injury. With these tools, healthcare professionals
possess the ability to both identify ‘high-risk’ populations, but more importantly identify
what biomechanical variables within these populations are deficient or malingering. The
ability to develop athlete-specific training protocols to target these malingering
biomechanical variables and reduce ACL injury risk is possible. Through this approach we
may be able to maximize the effects and reach of ACL injury prevention training protocols.
128
Again, from the ACL injury prevention framework proposed in chapter two, it is evident
there is a need to determine how biomechanically relevant risk factors like peak joint
loading and/or muscular support are influenced following a training intervention. From
chapters five and six we developed an open-source method to identify causal relationships
between an athlete’s kinematics and joint loading during complex, dynamic sporting tasks.
Using these methods we identified a generalised kinematic strategy to minimise valgus
knee loading and ACL injury risk during the weight acceptance phase of sidestepping; re-
position the whole body centre of mass medially, towards the desired change of direction
pathway. However, it should be recognized that these methods represent the first step of
many before these in-silico technique modifications can be recommended to community
level athletes. Future research is needed to determine if heterogeneous athletic
populations can be trained to use these in-silico technique recommendations to reduce
valgus knee loading and ACL injury risk in ‘ideal’ laboratory and then ‘real-world’ training
enviroments.
We also encourage future in-silico research to build upon these methods and findings.
For example, with a foot contact model, additional kinematic strategies to reduce valgus
knee moments during sidestepping and ACL injury risk may be established. Similar
research is also needed to determine if there exists a generalised kinematic strategy to
minimise valgus knee loading during the weight acceptance phase of single-leg landing. It
is with this type of information that we may be able to develop more effective ACL injury
prevention protocols and translate ACL focused research into injury prevention practice at
the community level.
Results from chapter three showed that conclusions can change when muscle activation
over a season of Australian football is analysed in isolation or in conjunction with changes
in knee loading. The role of the muscles in supporting the knee during sporting tasks like
sidestepping and single-leg landing should therefore not be overlooked. As the knee
flexes, the moment arms of muscles crossing the joint change, altering their ability to
support external knee loads (Lloyd and Buchanan, 2001). Further research is needed to
develop subject-specific models capable of quantifying the complex interaction between
lower limb kinematics, muscle force estimates, knee joint loading and ultimately ACL strain
to allow for a better assessment of how muscles function to support the knee and mitigate
ACL strain and injury risk during sporting tasks. It is with future research we may achieve
the ultimate goal of reducing ACL injury rates in the future.
129
Reference list chapter 7
Chappell, J.D., Limpisvasti, O., 2008. Effect of a neuromuscular training program on the kinetics and kinematics of jumping tasks. Am J Sports Med. 36 (6), 1081-1086.
Cochrane, J.L., Lloyd, D.G., Besier, T.F., Elliott, B.C., Doyle, T.L., Ackland, T.R., 2010. Training affects knee kinematics and kinetics in cutting maneuvers in sport. Med Sci Sports Exerc. 42 (8), 1535-1544.
Deci, E.I., Ryan, R.M., 1985. Intrinsic motivation and self-determination in human behaviour., ed. Plenum Press, New York. p. 11-39, 315-332.
Dempsey, A.R., Lloyd, D.G., Elliott, B.C., Steele, J.R., Munro, B.J., 2009. Changing sidestep cutting technique reduces knee valgus loading. Am J Sports Med. 37 (11), 2194-2200.
Finch, C., 2006. A new framework for research leading to sports injury prevention. J Sci Med Sport. 9 (1-2), 3-9; discussion 10.
Hewett, T.E., Stroupe, A.L., Nance, T.A., Noyes, F.R., 1996. Plyometric training in female athletes. Decreased impact forces and increased hamstring torques. Am J Sports Med. 24 (6), 765-773.
Lim, B.O., Lee, Y.S., Kim, J.G., An, K.O., Yoo, J., Kwon, Y.H., 2009. Effects of sports injury prevention training on the biomechanical risk factors of anterior cruciate ligament injury in high school female basketball players. Am J Sports Med. 37 (9), 1728-1734. Lloyd, D.G., Buchanan, T.S., 2001. Strategies of muscular support of varus and valgus isometric loads at the human knee. J Biomech. 34 (10), 1257-1267. Myer, G.D., Ford, K.R., McLean, S.G., Hewett, T.E., 2006. The effects of plyometric versus dynamic stabilization and balance training on lower extremity biomechanics. Am J Sports Med. 34 (3), 445-455. Myer, G.D., Ford, K.R., Palumbo, J.P., Hewett, T.E., 2005. Neuromuscular training improves performance and lower-extremity biomechanics in female athletes. J Strength Cond Res. 19 (1), 51-60. Twomey, D., Finch, C., Roediger, E., Lloyd, D.G., 2009. Preventing lower limb injuries: Is the latest evidence being translated into the football field? J Sci Med Sport. 12 (4), 452-456. Zebis, M.K., Bencke, J., Andersen, L.L., Dossing, S., Alkjaer, T., Magnusson, S.P., Kjaer, M., Aagaard, P., 2008. The effects of neuromuscular training on knee joint motor control during sidecutting in female elite soccer and handball players. Clin J Sport Med. 18 (4), 329-337.
130
APPENDIX A – TRAINING PROTOCOLS
A digital copy of the balance and technique training (Program 1) and ‘sham’ (Program 2)
training interventions used in chapters three and four can be found on the disk attached
with this thesis.
131
APPENDIX B – UWA UPPER AND LOWER BODY MODELS
A digital copy of the UWA upper and lower body models used in chapters three can be
found on the disk attached with this thesis.
132
APPENDIX C – SEMG ANALYSIS SOFTWARE
A digital copy of the surface electromyography software used to calculate the directed co-
contraction ratio’s and total muscle activation in chapter four can be found on the disk
attached with this thesis.
133
APPENDIX D – 37 DOF OPENSIM FULL BODY MODEL
A digital copy of the 37 DoF OpenSim full body model used in chapters five and six can be
found on the disk attached with this thesis.
134
APPENDIX E – OPENSIM KINEMATIC EXPORT SOFTWARE
A digital copy of the software used in chapters five and six to export the kinematics
recorded in Vicon into a file format recognized by OpenSim can be found on the disk
attached with this thesis.
135
APPENDIX F – OPENSIM GRF EXPORT SOFTWARE
A digital copy of the software used in chapters five and six to export ground reaction forces
recorded in Vicon into a file format recognized by OpenSim can be found on the disk
attached with this thesis.
136
APPENDIX G – OUTER-LEVEL OPTIMISATION SOFTWARE
A digital copy of the outer-level optimisation software used in chapters five and six can be
found on the disk attached with this thesis.
137
APPENDIX H – COMPARE FORCES PRE-TO-POST OPTMIZATION
A digital copy of the software used to compare a simulation’s knee loading pre-to-post
kinematic optimisation as per chapter four and five can be found on the disk attached with
this thesis.
138
APPENDIX I – COMPARE KINEMATICS PRE-TO-POST OPTMIZATION
A digital copy of the software used to compare a simulation’s full body kinematics and
identifying the critical joint coordinates from the kinetic maps as per chapter four and five
can be found on the disk attached with this thesis.
139
APPENDIX J – AVI IMAGES PRE TO POST OPTIMIZATION
Video files of a simulation pre-to-post kinematic optimisations as per chapter four and five
can be found on the disk attached with this thesis.