The Effect of Core Stability on Running Mechanics in Novice Runners

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Margaret Elisabeth Raabe

Graduate Program in Biomedical Engineering

The Ohio State University

2017

Dissertation Committee:

Ajit MW Chaudhari, PhD, Advisor

Alan S Litsky, MD, ScD

Robert A Siston, PhD

Thomas M Best, MD, PhD

Copyrighted by

Margaret Elisabeth Raabe

2017

Abstract

Despite the many health benefits associated with running, the annual running injury rate has been reported to be as high as 74%, and novice runners may be at the highest risk of developing these injuries. Research has shown core stability may affect lower extremity function, leading to the popular notion that insufficient core stabilization may lead to less efficient movements that ultimately contribute to musculoskeletal injury.

However, the role that core stability plays during running and its influence on injury risk is not well understood. The purpose of this dissertation was to establish the effect of core stability on fundamental mechanisms of running-related injuries and to investigate possible compensation strategies for reduced core stability.

Chapter 1 provides background information on running injuries, injury mechanisms, and core stability and describes the benefits of using dynamic simulations in combination with experimental data. Chapter 2 experimentally investigated the direct downstream effects of reduced core stability on running mechanics in novice runners and found reduced core stability was significantly associated with an increased external peak knee flexion moment (13.5±2.5 %BW*h vs 14.3±3.1 %BW*h, p=0.001) during the stance phase of running, which has previously been associated with increased patellofemoral joint loading. Chapter 3 describes the development and validation of an

OpenSim model that allows for the creation of simulations investigating full-body ii dynamics and contributions of the trunk muscles to dynamic tasks. In Chapters 4 and 5, the experimentally collected data from Chapter 2 was used with the model developed in

Chapter 3 to investigate the consequences of utilizing different possible compensation strategies for reduced core stability. Chapter 4 assessed the biomechanical consequences of altering running kinematics (kinematic compensation strategy) in response to reduced core stability and found this strategy was associated with increased internal knee loading during the stance phase of running (peak patellofemoral joint reaction force, p=0.029; knee abduction moment peak and impulse, p=0.01, p=0.02, respectively; peak knee extension moment, p=0.09), as well as reduced energy consumption (p=0.059), spinal loading (p≤0.06), and select peak core muscle forces (p≤0.06). Chapter 5 investigated utilizing a neuromuscular compensation strategy (altering only muscle activation strategies and maintaining kinematics) in response to core muscle fatigue and found this strategy was not associated with any change in estimated energy consumption or lower extremity loading during stance. Increased deep core muscle force production was observed as the only muscular compensation following core muscle fatigue, suggesting this may be the primary adjustment required to achieve a neuromuscular compensation strategy in the presence of core muscle fatigue. Therefore, insufficient core stability in novice runners may increase lower extremity loading and ultimately running injury risk.

A core neuromuscular training program emphasizing increased engagement and force production of the deep core muscles may give runners the ability to maintain movement patterns and utilize potentially lower-risk compensation strategies, such as a neuromuscular strategy, when core stability is compromised.

iii

Understanding how core stability affects running mechanics and potential compensation strategies used for poor core stability may ultimately contribute to the development of more effective and robust running injury prevention and rehabilitation regimens. The information presented in this dissertation improves the basic understanding regarding the influence of core stability on running mechanics in novice runners. This work will contribute to achieving the long-term goal of ultimately reducing the incidence of running-related injuries in novice runners.

Dedication

For my family --

Without you none of this would have been possible.

Acknowledgments

This research would not have been possible without the help and guidance of many people. I would first like to thank my advisor, Ajit Chaudhari, who provided endless guidance, mentorship, and support throughout all phases of this project. He has shaped me into the scientist I am today and I would not have made it this far without him.

I would also like to thank the other members of my dissertation committee; Alan Litsky,

Rob Siston and Tom Best. They each contributed a valuable perspective and expertise to this project, greatly improving the project’s quality and potential impact. Additionally, each committee member has been instrumental in my professional and personal growth over the past 5 years.

I am extremely grateful to have been a part of such a unique, interdisciplinary scientific environment in the Movement Analysis and Performance Laboratory and the

Sports Biomechanics Laboratory at Ohio State. Specifically, I would like to thank Jimmy

Oñate who has also provided support, guidance, and large contributions to many parts of this project; Scott Monfort, who has been by my side since day one, brainstorming, endlessly piloting protocols, collecting and analyzing data, troubleshooting, and providing instrumental feedback and collaboration; and Mike McNally, who was always been there to answer questions and provide support in any way possible. I would also like to thank Andrea Wanamaker, Greg Freisinger, Jackie Lewis, Chris Nagelli, Justin Creps, vi

Louise Thoma, Elena Caruthers, Sarah Schloemer, and everyone else in the lab for all the help and feedback they have provided along the way.

I would like to acknowledge the funding sources that supported multiple parts of this project: The Ohio State University’s Graduate School, The Ohio State University

Department of Biomedical Engineering, The National Institute of Arthritis and

Musculoskeletal and Skin Diseases, and The Ohio State University School of Health and

Rehabilitation Sciences.

Additionally, I would like to thank all my friends for providing support and encouragement during this process, as well as making sure I always kept some necessary balance in life. Specifically, Leiah, Allie and Erica, who all put up living with me during some of the most stressful times, and Greg, who has stood by me and has endlessly supported me even from hundreds of miles away.

Finally, and most importantly I would like to thank my family. Without them I would not have realized my passion for math, science and engineering, and would not have had the support to pursue this journey. I am extremely thankful for their unconditional love and encouragement, and the superb values and work ethic they instilled in me. It is only because of them that any of this was possible.

vii

Vita

2008 ...... High School Diploma, Bexley High School

2012 ...... B.A. Physics, The College of Wooster

2015 ...... M.S. Biomedical Engineering, The Ohio

State University

2012 to present ...... Graduate Research Associate, The Ohio

State University

Publications

Raabe, M. E. and A. M. W. Chaudhari (2016). "An Investigation of Jogging Biomechanics using the Full-Body Lumbar Spine Model: Model Development and Validation." Journal of Biomechanics 49: 1238-1243.

Fields of Study

Major Field: Biomedical Engineering

viii

Table of Contents

Abstract ...... ii

Dedication ...... v

Acknowledgments...... vi

Vita ...... viii

Publications ...... viii

Fields of Study ...... viii

Table of Contents ...... ix

List of Tables ...... xiv

List of Figures ...... xvii

List of Abbreviations ...... xxiii

Chapter 1 : Introduction ...... 1

1.1 Running Injuries ...... 1

1.1.1 Background and Prevalence ...... 1

1.1.2 Common Running Injuries ...... 2

1.1.3 Biomechanical Injury Mechanisms ...... 6

1.2 Core Stability...... 8

1.2.1 Background and Definition ...... 8

1.2.2 Core Stability and Lower Extremity Function ...... 10

1.2.3 Core Stability and Lower Extremity Injuries ...... 11

1.2.4 Lack of Evidence Linking Core Stability and Running Injuries ...... 12

1.2.5 Compensation Strategies for Poor Core Stability are Unknown ...... 14

1.3 The Power of Dynamic Simulations of Human Movement ...... 15

1.3.1 Dynamic Simulations Complement Experiments Well ...... 15

1.3.2 OpenSim and its Capabilities ...... 16

1.4 Statement of Purpose ...... 19

1.5 Outline of Upcoming Chapters ...... 20

Chapter 2 : An Experimental Investigation of the Direct Downstream Effects of Reduced

Core Stability on Running Mechanics in Novice Runners ...... 22

2.1 Abstract ...... 22

2.2 Introduction ...... 23

2.3 Methods ...... 25

2.3.1 Participants ...... 25

2.3.2 Experimental Procedures ...... 25

2.3.3 Statistics ...... 32

2.4 Results ...... 33

2.4.1 Muscle Fatigue and Core Stability ...... 33

2.4.2 Biomechanical Running Parameters ...... 35

2.5 Discussion ...... 38

2.5.1 Implications on Injury Risk in Novice Runners ...... 38

2.5.2 Limitations ...... 42

2.6 Conclusion ...... 43

2.7 Acknowledgements ...... 43

Chapter 3 : An Investigation of Jogging Biomechanics using the Full-Body Lumbar Spine

Model: Model Development and Validation ...... 44

3.1 Abstract ...... 44

3.2 Introduction ...... 45

3.3 Methods ...... 46

3.3.1 Model Development ...... 46

3.3.2 Model Validation ...... 49

3.4 Results and Discussion ...... 52

3.4.1 Model Parameters ...... 52

3.4.2 Muscle function ...... 54

3.4.3 Simulations ...... 57

3.4.4 Limitations ...... 61

3.5 Conclusion ...... 62

3.6 Acknowledgements ...... 62

Chapter 4 : The Biomechanical Effects of Utilizing a Kinematic Compensation Strategy in Response to Reduced Core Stability during Running ...... 63

4.1 Abstract ...... 63

4.2 Introduction ...... 64

4.3 Methods ...... 68

4.3.1 Participants ...... 68

4.3.2 Core Stability Knockdown and Running Biomechanics ...... 68

4.3.3 Simulation procedures ...... 71

4.3.4 Statistics ...... 75

4.4 Results ...... 78

4.4.1 Kinematic Changes ...... 78

4.4.2 Effect of a Kinematic Compensation Strategy ...... 79

4.5 Discussion ...... 87

4.5.1 Limitations ...... 90

4.6 Conclusion ...... 91

Chapter 5 : Simulating the Utilization of a Neuromuscular Compensation Strategy in

Response to Core Muscle Fatigue during Running ...... 92

5.1 Abstract ...... 92

5.2 Introduction ...... 93

5.3 Methods ...... 97

5.3.1 Participants ...... 97

5.3.2 Core Stability Knockdown and Running Biomechanics ...... 97

5.3.3 Simulation procedures ...... 100

5.3.4 Statistics ...... 102

5.4 Results ...... 104

5.5 Discussion ...... 113

5.5.1 Limitations ...... 116

5.6 Conclusion ...... 118

xii

Chapter 6 : Research Findings and Future Work ...... 119

6.1 Contributions ...... 119

6.2 Future Work ...... 126

6.3 Summary ...... 129

Appendix A : Supplemental Material for Chapter 2 ...... 133

Appendix B : Supplemental Methods and Results for the Development and Validation of the Full-Body Lumbar Spine (FBLS) Model ...... 135

B.1 Supplemental Methods ...... 135

B.1.1 Model Development...... 135

B.1.2 Altered Maximum Isometric Forces ...... 137

B.2 Supplemental Results ...... 140

B.2.1 Full Table of Trunk Muscle Moment Arms ...... 140

B.3 Simulated Muscle Forces...... 146

B.3.1 Standing ...... 146

B.3.2 Running ...... 149

Appendix C : Full Linear Mixed Model Effects for Chapters 4 and 5 ...... 152

Appendix D : Individual Participant Data for Subject-Specific Simulations ...... 160

Appendix E : Residual and Reserve Results for OpenSim Simulations ...... 162

References ...... 195

xiii

List of Tables

Table 2.1.Core Stability Knockdown Protocol (CSKP) designed to fatigue both the superficial and deep core musculature...... 26

N=10)...... 37

Table 2.3. Mean change ± standard error (SE) from pre- to post-CSKP and the p-values corresponding to the CSKP mixed-model effect for the biomechanical variables found to be significant in the reduced core stability (CS) group compared to the changes seen in the non-core stability (NCS) group...... 38

Table 3.1. Sagittal plane moment arms for each of the RA, EO, and IO muscle group fascicles...... 54

Table 4.1. Peak joint angles during running before and after the core stability knockdown protocol (CSKP) for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group)...... 79

xiv

Table 4.3. Differences in spinal loading variables from the pre-CSKP kinematic state to the post-CSKP kinematic state and the corresponding mixed model kinematic effect p- values...... 85

Table 4.4. Differences in peak muscle force variables from the pre-CSKP kinematic state to the post-CSKP kinematic state and the corresponding mixed model kinematic effect p- values...... 87

Table 5.1. Differences in spinal loading variables from the baseline (no core muscle fatigue) state to the fatigued state and the corresponding mixed model fatigue effect p- values...... 106

Table A.1. Change from pre- to post-CSKP and mixed model effects (slopes) for each of the biomechanical parameters...... 134

Table B.1. Changes made to the maximum isometric force property (Fm0) in the FBLS model for some lower extremity and trunk muscles and the scaling factor that was applied...... 138

Table B.2. Sagittal plane moment arms for all trunk muscle fascicles...... 141

Table B.3. Comparison of compared trunk muscle forces (N) simulated during standing to trunk muscle force predictions previously published in the literature (Schultz,

Haderspeck et al. 1983; Calisse, Rohlmann et al. 1999; Wilke, Rohlmann et al. 2003;

Rohlmann, Bauer et al. 2006)...... 148

Table C.1. All results from linear mixed-models for estimated energy consumption and lower extremity loading variables...... 153

Table C.2. Results from linear mixed-models for spinal compressive and anterior shear loading variables...... 154

Table C.3. Linear mixed-model results for muscle force variables...... 157

Table D.1. Individual participant data for subject-specific simulations...... 161

xvi

List of Figures

Figure 1.1. Simplified diagram of forces acting at the knee during knee flexion...... 4

Figure 1.2. Anatomy of the iliotibial band (ITB). The ITB is a band of dense connective tissue that attaches proximally to the pelvis and distally to Gerdy’s tubercle on the anterolateral aspect of the tibia. ITB syndrome is associated with sharp pain felt on the lateral aspect of the knee (red circle). Image adapted from Fredericson & Tenforde...... 5

Figure 1.3. Core stability is dependent on multiple factors that interact in order to create a stable system...... 10

Figure 1.4. OpenSim workflow for jogging simulations, adapted from Delp et al. 2007. 17

Figure 1.5. Dissertation framework and outline...... 20

Figure 2.1. Experimental modified point-cluster set used for movement analysis...... 28

Figure 2.2. Unstable quiet sitting task (QST) used to quantify core stability...... 31

Figure 2.4. Schematic of study results and their expected effect on patellofemoral contact mechanics...... 40

Figure 3.1. The full-body lumbar spine (FBLS) model...... 47

xvii

Figure 3.2. Maximum isometric joint moments for axial rotation (A), lateral bending (B), trunk extension (C), and trunk flexion (D) in the model compared to experimental data collected in this study and data in the literature examining trunk strength at multiple joint angles...... 56

Figure 3.3. Trunk and lower extremity jogging kinematics simulated using the FBLS model over one right foot gait cycle...... 58

Figure 3.4. Trunk and lower extremity (LE) joint moments during jogging simulated by

Hamner et al 2010 and using the FBLS model...... 59

Figure 3.5. EMG for all collected muscles (dashed line) compared to model activations estimated in OpenSim (solid line) using Static Optimization...... 60

Figure 4.1. Diagram of the study’s purpose (A), hypotheses (B), and motivation (C). .... 67

Figure 4.2. Schematic describing the four subject-specific simulations developed for each participant...... 73

Figure 4.3. Previously recorded data investigating how a muscle’s change in force production is related to a change in its MedF...... 74

Figure 4.4. Diagram of the spinal column showing the vertebral loads analyzed in this study...... 77

Figure 4.6. Individual changes in the ipsilateral patellofemoral joint reaction force (JRF) during stance from pre- to post-CSKP in the CS group in units of body weight (BW). .. 84

xviii

Figure 5.1. Diagram of the study’s purpose (A), hypotheses (B), and motivation (C). .... 96

Figure 5.2. Deep core muscles utilized as compensators in response to core muscle fatigue during running...... 109

Figure 5.4. Individual changes in peak force production of the deep core muscles in the

CS group acting as compensators in response to a subject-specific level of overall core muscle fatigue...... 112

Figure B.1. Estimated internal axial compressive loads on each of the lumbar vertebrae during standing compared to loads previous published in the literature (Nachemson and

Morris 1964; Callaghan and McGill 2001; El-Rich, Shirazi-Adl et al. 2004)...... 146

Figure B.2. Trunk and lower extremity muscle forces during jogging estimated using the

FBLS model compared to those previously published using a more simplified full-body model (Hamner, Seth et al. 2010)...... 149

Figure E.2. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 1...... 163

Figure E.3. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 2...... 164

Figure E.4. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 3...... 165

xix

Figure E.5. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 4...... 166

Figure E.6. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 1...... 167

Figure E.7. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 2...... 168

Figure E.8. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 3...... 169

Figure E.9. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 4...... 170

Figure E.10. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 1...... 171

Figure E.11. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 2...... 172

Figure E.12. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 3...... 173

Figure E.13. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 4...... 174

Figure E.14. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 1...... 175

Figure E.15. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 2...... 176

Figure E.16. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 3...... 177

Figure E.17. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 4...... 178

Figure E.18. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 1...... 179

Figure E.19. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 2...... 180

Figure E.20. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 3...... 181

Figure E.21. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 4...... 182

Figure E.22. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 1...... 183

Figure E.23. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 2...... 184

Figure E.24. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 3...... 185

Figure E.25. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 4...... 186

Figure E.26. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 1...... 187

xxi

Figure E.27. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 2...... 188

Figure E.28. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 3...... 189

Figure E.29. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 4...... 190

Figure E.30. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 1...... 191

Figure E.31. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 2...... 192

Figure E.32. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 3...... 193

Figure E.33. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 4...... 194

xxii

List of Abbreviations

BW: body weight GRF: ground reaction force

BF: biceps femoris h: height

BMI: body mass index HAbd: hip abduction

CMC: computed muscle control HAdd: hip adduction

CoPexc: center of pressure excursion HIR: hip internal rotation

CS group: participants that experienced IAP: intra-abdominal pressure a meaningful reduction in core stability IK: inverse kinematics post-CSKP IL_L: lumbar component of iliocostalis

CSKP: core stability knockdown lumborum protocol IL_R: thoracic/rib component of

Dom: dominant-side iliocostalis lumborum

EMG: electromyography IO: internal obliques

EO: external obliques ITB: iliotibial band

ES: erector spinae ITBS: iliotibial band syndrome

FBLS: Full-Body Lumbar Spine JRF: joint reaction force

m F 0: maximum isometric force KAbd: knee abduction

GMax: gluteus maximus KAdd: knee adduction

GMed: gluteus medius KF: knee flexion

GMin: gluteus minimus KIR: knee internal rotation

xxiii

L5: lumbar extensor at the level of the PS: psoas major

th 5 lumbar vertebrae pTibcomp: peak tibial compression

LD: latissimus dorsi QL: quadratus lumborum

LE: lower extremity QST: quiet sitting test

LG: lateral gastrocnemius RA: rectus abdominis

LTpL: lumbar component of longissimus RF: rectus femoris thoracis RRA: residual reduction algorithm

LTpT: thoracic component of SO: static optimization longissimus thoracis Sol: soleus

MedF: median frequency ST: semitendinosus

MF: multifidus TA: tibialis anterior

NDom: non-dominant side TrA: transverse abdominis

NCS group: participants with no change TSF: tibial stress fracture in core stability post-CSKP VALR: vertical GRF average loading

PFJ: patellofemoral joint rate pFPFJ: peak patellofemoral joint reaction VL: vastus lateralis force VM: vastus medialis

PFPS: patellofemoral pain syndrome

xxiv

Chapter 1 : Introduction

1.1 Running Injuries

1.1.1 Background and Prevalence

Running is a sports activity that continues to grow in popularity each year. In

2014, almost 19 million people completed a running road race in the USA, which is a

300% increase from 1990 (Running 2015). Running is an excellent exercise for people who want to maintain or increase their fitness level in order to avoid health problems associated with physical inactivity because it can provide quick health benefits, is inexpensive, requires no special skills to perform, and may even provide social interaction. With the number of novice and recreational runners increasing, running seems to show potential to improve public health. However, healthy running habits are unfortunately frequently interrupted by running injuries. The annual running injury rate has been reported to range from 24-74% (Marti, Vader et al. 1988; Macera, Powell et al.

1989; van Mechelen 1992; Buist, Bredeweg et al. 2010; Daoud, Geissler et al. 2012) and despite many improvements in footwear and other technology over the past few decades, these injury rates have yet to decline (Daoud, Geissler et al. 2012; Rixe, Gallo et al.

2012).

Novice runners are believed to be the most susceptible to developing running injuries, possibly due to a lack of training experience (Buist, Bredeweg et al. 2010;

Tonoli, Cumps et al. 2010; Verhagen 2012; Schmitz, Russo et al. 2014). These injuries 1 are detrimental to public health because they are associated with high socio-economic costs (Buford, Ivey et al. ; Buff, Jones et al. 1988) and can influence the physically active future of an individual greatly. In novice runners, development of a running-related injury during training has been associated with discontinuing a training regimen even after the injury has healed (Buist, Bredeweg et al. 2010), and the failure of building a physically active lifestyle (Sallis, Hovell et al. 1992). Literature identifying prevention strategies for running related injuries is scarce and most intervention studies implementing currently available injury prevention products or programs have been unsuccessful (Buist,

Bredeweg et al. 2008; Yeung, Yeung et al. 2011). Reducing the rate of running related injuries, especially in novice runners, has the potential to improve both short-term and long-term public health.

1.1.2 Common Running Injuries

With each step taken during running, the lower leg and knee can experience forces up to 11 times a person’s body weight (Scott and Winter 1990). Because these high loads are repeated over hundreds of strides during a typical run, the majority of running injuries are overuse injuries that result from cumulative microtrauma to a structure. Some of the most common running injuries include patellofemoral pain syndrome (PFPS), iliotibial band syndrome (ITBS), and tibial stress fractures (TSF)

(Taunton, Ryan et al. 2002).

The patella is a sesamoid bone located anterior to the knee joint, sitting slightly above or within the trochlear groove of the femur depending on the angle of knee flexion.

It serves as the central attachment point for the quadriceps tendon and the patellar

2 ligament. The patella plays a large role in running, as its main function is to increase the mechanical advantage of the knee extensor muscles. The patella displaces the quadriceps tendon line of action away from the joint, increasing the moment arm of the quadriceps force, allowing the muscles to generate a larger torque for a given amount of muscle force (Fox, Wanivenhaus et al. 2012). As the knee flexes during running, force is generated in both the quadriceps tendon and the patellar ligament and the patella is subjected to a compressive patellofemoral joint (PFJ) force (Fig. 1.1). PFPS is the most common running injury and is characterized by aching pain behind and around the patella

(Taunton, Ryan et al. 2002; Ferber, Bolgla et al. 2015). Over-loading of the patella and patellar maltracking resulting from atypical running kinematics are the primary injury mechanisms that have been associated with PFPS (Lee, Anzel et al. 1994; James 1995;

Mizuno, Kumagai et al. 2001; Heino Brechter and Powers 2002; Chen and Powers 2014).

If the patella is not moving properly within the trochlear groove as the knee flexes and extends this may result in PFJ loads applied over a reduced contact area leading to increased PFJ stress, which ultimately may lead to the pain associated with PFPS

(Mizuno, Kumagai et al. 2001; Willy, Manal et al. 2012). Heino Brechter and Powers used a biomechanical model to compare patellofemoral joint stress during walking between participants with and without PFPS and found the PFPS group had over 2x greater peak stress on the patella than the control group (3.13 MPa vs 6.61 MPa) (Heino

Brechter and Powers 2002).

Figure 1.1. Simplified diagram of forces acting at the knee during knee flexion. Abbreviations: force in quadriceps (Fq), force in patellar ligament (Fplig), compressive patellofemoral force (Fpf), external knee flexion moment (Mkf).

The iliotibial band (ITB) is a band of dense connective tissue spanning the lateral aspect of the thigh (Fig. 1.2) (Fredericson and Tenforde 2016). The ITB attaches proximally to the pelvis and distally to the lateral condyle of the tibia (Gerdy’s tubercle).

Some research has also shown the ITB may attach distally to the lateral femoral epicondyle as well ( Fairclough, Hayashi et al. 2006). The primary function of the ITB during running is to stabilize the knee and prevent the hip from excessively adducting.

ITBS is the most common cause of lateral knee pain (Noble 1980; Ferber, Noehren et al.

2010) and is associated with a sharp pain on the lateral aspect of the knee near the ITB’s attachment to the lateral femoral epicondyle (Fig. 1.2). There are two primary injury mechanisms believed may cause ITBS. The first is friction that arises between the ITB

4 and the lateral femoral epicondyle when the knee flexes and extends during running and the ITB passes over the epicondyle (Orchard, Fricker et al. 1996). This friction may cause inflammation of the ITB (Lavine 2010) and consequently the pain associated with ITBS.

The second injury mechanism is elevated stress placed on the ITB when it is excessively stretched during atypical movements, such as excessive hip adduction or knee internal rotation, which may cause compression of a loose layer of fat and connective tissue that lies under the ITB near its attachment to the femoral epicondyle (Andriacchi, Kramer et al. 1985; James 1995; Fairclough, Hayashi et al. 2006; Ferber, Noehren et al. 2010).

Both of these proposed injury mechanisms involve increased stress and tension in the

ITB as the direct mechanical cause of ITBS.

The tibia is the weight bearing bone of the lower leg and during running the compressive loads experienced by the tibia have been estimated to range from 10-14 times a person’s body weight (Scott and Winter 1990). When a bone experiences low levels of repetitive stress with an adequate amount of recovery time, Wolff’s Law states that it can remodel and adapt to the stress. However, when bones experience repetitive loads at high magnitudes and/or at high loading rates, stress to the bone accumulates and the bone may be unable to repair and remodel quickly enough in response to this loading

(Harrast and Colonno 2010). Eventually without sufficient time to remodel, this increase in stress will ultimately lead a partial or incomplete fracture of the bone, referred to as a stress fracture (Romani, Gieck et al. 2002). Tibial stress fractures (TSF) are the most common site of stress fractures in runners (Matheson, Clement et al. 1987). A TSF is generally characterized by localized pain and tenderness of the tibia during or following physical activity (Harrast and Colonno 2010). The primary injury mechanism for TSFs is ultimately over-loading of the tibia during running resulting from excessive loading magnitudes, loading frequency, and loading rates (Milner, Ferber et al. 2006; Pohl,

Mullineaux et al. 2008; Zadpoor and Nikooyan 2011; Davis, Bowser et al. 2015).

1.1.3 Biomechanical Injury Mechanisms

A number of previous scientific studies have identified biomechanical risk factors associated with common running injuries. Abnormal lower extremity movement during running has been identified as one primary mechanism for many running injuries. A two- year prospective study with female runners found those that developed ITBS exhibited larger hip adduction and knee internal rotation angles at foot strike compared to those that

6 were uninjured (Noehren, Davis et al. 2007), both of which may increase the tension in the ITB and increase its compression against the lateral femoral epicondyle. A similar group of female runners was also followed prospectively for two years and it was found that those that developed PFPS exhibited a greater peak hip adduction angle during stance than those that remained injury-free (Noehren, Hamill et al. 2013), which may result in increased lateral patellar contact stress (Huberti and Hayes 1984). Additionally, a cadaveric study showed that increased femoral internal rotation resulted in increased contact pressure on the lateral facet of the patella (Lee, Anzel et al. 1994). Since it has been shown using magnetic resonance imaging that symptomatic PFPS patients have more cartilage damage to the lateral aspect of the patella (Farrokhi, Colletti et al. 2011), this suggests that excessive hip internal rotation during running may also be a risk factor for PFPS. Lastly, it is believed excessive knee abduction may increase risk of developing

PFPS, as increased knee abduction may increase the lateral force that acts on the patella

(Powers 2003) and has been shown to alter patellar tracking in such a way that may increase lateral patellofemoral contact pressures (Mizuno, Kumagai et al. 2001).

Dynamic over-loading of the lower extremities during running has also previously been associated with many running injuries. Runners with a history of TSF were found to exhibit increased vertical ground reaction force (GRF) loading rates during the impact phase of stance and greater peak tibial acceleration when compared to healthy controls

(Milner, Ferber et al. 2006). Additionally, a large group of runners was followed prospectively for two years and those who were medically diagnosed with injuries had larger vertical GRF impact peaks and average vertical GRF loading rates than those

7 remained uninjured (Davis, Bowser et al. 2015). In a group of prospectively followed marathon runners, those that became injured exhibited significantly increased external peak hip adduction moments and impulses (MacMahon, Chaudhari et al. 2000).

Additionally, during a one-legged standing task with healthy individuals, Tateuchi et al. found that standing posture with increased contralateral trunk lean and pelvis tilt was associated with an increased external hip adduction moment and increased ITB hardness, suggesting an elevation in this moment may increase tension in the ITB (Tateuchi,

Shiratori et al. 2015). Furthermore, Stefanyshyn et al. prospectively followed runners throughout a summer training season and found that runners who developed PFP had higher internal knee abduction angular impulses than matched patients who remained uninjured (Stefanyshyn, Stergiou et al. 2006), a biomechanical change that may be indicative of increased stress on the ITB (James 1995). An increased external knee flexion moment is also believed to increase PFPS risk, as this moment during stance must be balanced by the quadriceps and an increase in this moment likely leads to a direct increase in tension in the quadriceps tendon and patellar ligament and consequently, larger forces in the PFJ (Andriacchi, Kramer et al. 1985; James 1995).

1.2 Core Stability

1.2.1 Background and Definition

In recent years, core stability has become an extremely popular topic among physical therapists, athletic trainers, clinicians, researchers, and physically active people in general, especially runners (Kibler, Press et al. 2006; McGill 2010). For the purpose of this study, the core is defined as the region of the body bounded by the pelvis and

8 diaphragm including the muscles of the abdomen and lower back. The muscles that cross the hip joint are often also considered part of the core; however, their primary function during running is to move the limbs. Since the primary function of the hip muscles during running is fundamentally different from that of the muscles of the abdomen and lower back, which act primarily to move or hold the torso, they are excluded from our definition of the core.

Core stability will be defined as the body’s ability to maintain or resume an equilibrium trunk and pelvis position (or trajectory) following internal or external perturbations (Zazulak, Hewett et al. 2007). During running these perturbations may come from the body itself, or from environmental surroundings. Both passive and active components of the trunk contribute to core stability (Panjabi 1992), but it has been shown that the passive elements cannot stabilize the spine on their own during activities of daily living (Crisco, Panjabi et al. 1992). Therefore, the active core musculature is crucial to maintaining core stability, and will be focused on in this project.

Core stability is a dynamic property, as the core musculature must constantly react and adapt to varying positions and loading conditions the body experiences during running (Willson, Dougherty et al. 2005). One’s ability to stabilize the core is comprised of multiple components that interact to create a stable system. These components include core muscle strength, endurance, and neuromuscular coordination or control (Figure 1.3).

Figure 1.3. Core stability is dependent on multiple factors that interact in order to create a stable system. These include muscle strength, muscle endurance, and neuromuscular coordination of the core musculature.

1.2.2 Core Stability and Lower Extremity Function

Since over half of the mass of the body resides in the upper body (Dempster and

Gaughran 1967), it is likely that control of this mass will have a large influence on the function of the lower extremities and the loads they experience. In 1991, Bouisset proposed the idea that stabilization of the trunk and pelvis is necessary for, and must occur prior to, all extremity movements (Bouisset 1991). Since then, research has confirmed that muscle activity of the trunk precedes voluntary movement of the lower extremities (Hodges and Richardson 1997), from which it was concluded that the abdominal muscles contract to stabilize the spine and provide a stable foundation for effective and efficient production, absorption, and control of force and motion to the

10 lower extremities. These conclusions have consequently led to the popular notion that insufficient core stability may lead to movements that are less efficient and ultimately lead to injury (Willson, Dougherty et al. 2005), however this theory has been supported with minimal scientific evidence.

1.2.3 Core Stability and Lower Extremity Injuries

Poor core stability has been associated with incidence of some lower extremity injuries. A prospective study followed collegiate athletes for three years and found poor trunk control, measured using an isolated trunk perturbation test, was associated with

ACL injury incidence (Zazulak, Hewett et al. 2007). The same population also had core proprioception measured as their ability to accurately return to the initial trunk position after being rotated in the transverse plane (Zazulak, Hewett et al. 2007). It was found that women who experienced knee and ligament/meniscal injuries over the three year period exhibited significant deficits in active core proprioceptive repositioning (Zazulak,

Hewett et al. 2007).

A number of studies have also provided evidence suggesting that improper positioning of the trunk may increase knee loading associated with ACL injuries. Lateral trunk lean during a run-to-cut maneuver was positively associated with peak knee abduction moment, suggesting increased torso lean may increase ACL risk, as an increase in the knee abduction moment has been associated with increased ACL strain and injury incidence (Jamison, Pan et al. 2012). Dempsey et al. found that healthy athletes who intentionally increased their trunk lean and rotation during a 45° side-step cutting maneuver, experienced increased loading of the knee previously associated with

11 increased ACL injury risk (Dempsey, Lloyd et al. 2007). Donnelly et al. used the musculoskeletal modeling software OpenSim to optimize experimentally collected movement data during an unanticipated 45° cutting maneuver and found that the optimal adjustments to whole-body movements, defined as ones that minimized the peak knee abduction moment during the cutting task, involved repositioning of the trunk and the whole body center of mass to be closer to the midline (Donnelly, Lloyd et al. 2012).

Ankle sprains have also been associated with poor control of the trunk. Bullock-

Saxton el al. found people with a history of severe ankle sprains exhibited delayed activation onset of the hip extensor muscles during a hip extension task compared to healthy controls, suggesting ankle sprains may be linked with deficiencies in neuromuscular control of the lumbopelvic-hip complex (Bullock-Saxton, Janda et al.

1994).

1.2.4 Lack of Evidence Linking Core Stability and Running Injuries

The majority of scientific studies that have currently examined the relationship between core stability and running injuries are cross-sectional or retrospective. As a result, it is not possible to determine whether the injuries studied were a cause or an effect of reduced core stability. Additionally, many of these studies defined the core musculature to include the trunk and hip musculature and primarily focused on the muscular strength or endurance component of core stability. A number of studies have found injured runners exhibit less strength in the hip musculature of the injured limb when compared to the uninjured limb, as well as compared to healthy controls

(Fredericson, Cookingham et al. 2000; Cichanowski, Schmitt et al. 2007; Dierks, Manal

12 et al. 2008; Ferber, Hreljac et al. 2009). For example, a study comparing females that currently had PFPS to healthy controls found the injured group had significantly less hip abduction and external rotation strength than healthy controls (Ireland, Willson et al.

2003). Similarly, another study comparing a group of runners with ITBS to healthy controls found that those suffering ITBS had significantly less hip abduction strength in the involved limb (Fredericson, Cookingham et al. 2000). After putting the injured runners through a 6-week hip abductor strengthening program 92% of the injured group was pain free (Fredericson, Cookingham et al. 2000). Ferber et al. compared two 6-week rehabilitation protocols, one incorporating hip and core muscle strengthening and the other being a more standard knee-focused program, performed by runners currently suffering from PFP and found those participating in the hip and core focused rehabilitation program had an earlier resolution of pain and increased hip muscle strength compared to the knee-focused group (Ferber, Bolgla et al. 2015).

A limited number of studies have examined the relationship between the neuromuscular control aspect of core stability and running injuries. Many of these studies also include the hip musculature in their definition of the core. Jamison et al. investigated correlations between pelvic control and running mechanics and found increased pelvic control was associated with biomechanical changes that have been associated with decreased ITBS, PFPS and TSF injury risk (Jamison and Chaudhari 2009). Shirazi et al. compared how a group of healthy subjects and a group with PFPS responded to an unexpected lateral perturbation applied to the pelvis (Shirazi, Moghaddam et al. 2014).

They found the injured group utilized different core muscle recruitment patterns than the

13 healthy group, in which the lumbar extensors of the injured group were activated earlier and longer and the gluteus medius activation had a significantly delayed onset. Cowan et al. also showed impairment in the neuromuscular control of the gluteus medius muscle in patients with PFPS, as they found activation of this muscle was delayed during stair climbing when compared to healthy controls (Cowan, Crossley et al. 2009).

The influence of one’s ability to control the core musculature on running injury risk remains not well understood. Studies from which causal conclusions can be drawn and that incorporate the multiple components of core stability are needed to further understand the role that core stability plays in running and the influence of core stability on running injury risk. Additionally, isolating the contribution of the trunk musculature to core stability during running from that of the hip musculature may be beneficial to better comprehend the influence of the lumbopelvic complex on lower extremity loading and injury risk.

1.2.5 Compensation Strategies for Poor Core Stability are Unknown

To our knowledge, there have been no scientific studies that identify and compare the different ways that individuals compensate for poor core stability. A better understanding of what strategies may be utilized to compensate for poor core stability will likely lead to improved identification of runners lacking core stability and further the knowledge of what injuries may arise in runners with this deficiency. Additionally, this information may ultimately contribute to the development of improved injury prevention and rehabilitation protocols. New research paradigms are needed to isolate and identify

14 compensation strategies that may be utilized by runners when strength and the other aspects of core stability are impaired in real-life situations.

1.3 The Power of Dynamic Simulations of Human Movement

1.3.1 Dynamic Simulations Complement Experiments Well

Dynamic computer simulations are an excellent complement to experiments.

Simulations further the understanding of movement dynamics by enabling the measurement of important variables that cannot be measured experimentally, including forces generated by individual muscles and joint contact forces. Measuring individual muscle forces and internal joint loads experimentally would require extremely difficult and invasive techniques. Additionally, in an experimental laboratory environment it is nearly impossible to change only one variable and hold all others constant. Simulations allow cause-effect relationships in complex dynamic systems to be established and “what if?” studies to be performed where one variable is changed and the rest are held constant.

OpenSim is a powerful freely available, open-source software system used by researchers around the world to develop musculoskeletal models and create dynamic simulations of many different movements (Delp, Anderson et al. 2007). Simulations allow for an improved understanding of muscle activation patterns (van der Krogt, Delp et al. 2012), contributions of individual muscles to gait (Hamner, Seth et al. 2010), internal joint loading (Lenhart, Thelen et al. 2014), and even energy consumption during dynamic movements (Umberger, Gerritsen et al. 2003; Umberger 2010).

1.3.2 OpenSim and its Capabilities

OpenSim is a simulation software with a number of tools that enable modeling, simulation, and analysis of the neuromusculoskeletal system (Delp, Anderson et al.

2007). A public repository containing data, models, and computational tools that relate to physics-based simulation is available for all OpenSim users to utilize and build-upon for their specific research needs (Delp, Anderson et al. 2007). In OpenSim, experimentally collected kinematic and kinetic data can be imported and used as input to create kinematically-driven simulations of dynamic movement. From a simulation, the individual muscle forces and internal joint loads during the movement can be estimated and analyzed. Additionally, in OpenSim individual muscle parameters can be altered (e.g. maximum isometric force, tendon slack length, optimal fiber length) and a simulation can then be repeated to examine the direct consequences of the change in a given parameter on output variables such as internal joint loads, muscle forces, and estimated energy consumption during running.

Figure 1.4 shows an example of the workflow and tools used in OpenSim that can be used to develop subject-specific dynamic simulations which output muscle activations, forces, and internal joint loads. In this workflow, a generic musculoskeletal model is first scaled to anthropometrically match the individual participant for whom data had been collected (Delp, Anderson et al. 2007). Next, motion and mass properties of the model are optimized using inverse kinematics (IK) and a residual reduction algorithm (RRA) to achieve a dynamically consistent set of kinematics and kinetics that best match the

16 experimentally collected data (Delp, Anderson et al. 2007). Next, static optimization

(SO) is performed to resolve the net joint moments into individual muscle forces at each instant in time (Steele, Demers et al. 2012). Lastly, JointReaction Analysis is then performed using the Analyze tool to calculate the internal joint loads during a given task

(Steele, Demers et al. 2012).

Figure 1.4. OpenSim workflow for jogging simulations, adapted from Delp et al. 2007.

Very few studies have used OpenSim to create dynamic simulations of running gait (Hamner, Seth et al. 2010; Arnold, Hamner et al. 2013; Creps 2014) and to our knowledge, only one has examined the role that the core plays in running (Creps 2014).

This is likely largely due to the fact that prior to this study, the only available full-body

OpenSim musculoskeletal models had over-simplified torsos, with the entire spine modeled as one rigid segment and very few trunk muscles. This limitation prevented meaningful simulation investigations of the relationships between core stability and lower-extremity loading during running and restricted the translation of simulation results to real-world situations.

Although simulations provide many benefits for biomechanical investigations, there are limitations of biological simulations that must be considered when interpreting results from simulation studies. First, the ability of simulation results to translate to real- life situations is largely dependent on the physiological accuracy of the musculoskeletal model and how well the given model describes the participant population being investigated. Additionally, assumptions made during the simulation process must be considered. For example, in SO muscle forces are resolved and distributed across a joint using an optimization criterion which minimizes the sum of squared activations during a dynamic task. This objective function assumes the goal of the neural system during a given task is to minimize muscle activity. Lastly, utilizing SO rather than a forward- driven tool like Computed Muscle Control (CMC) in the simulation workflow minimizes computational cost; however, SO does not take into account activation dynamics and is also a frame-by-frame solver, rather than integrating forward in time like CMC so anticipatory activations that may occur during a task (i.e., in preparation for heel strike) are likely lost in SO (Thelen and Anderson 2006; Hicks and Dembia 2014).Continuing to improve the physiological accuracy of musculoskeletal models in OpenSim, especially in

18 regards to the torso region, will expand the set of possible “what if” simulation scenarios and increase our ability to interpret the role core stability plays in running.

1.4 Statement of Purpose

Running-related injury rates remain very high despite many improvements in technology and growing body of research over the past few decades (Daoud, Geissler et al. 2012; Rixe, Gallo et al. 2012). Taking into account the growing popularity of running each year and its associated health benefits, reducing the rate of running-related injuries, especially in novice runners, has the potential to improve both short-term and long-term public health. The main purpose of this research was to establish the effect of core stability on the fundamental biomechanical mechanisms of running injuries that have previously been identified to influence running injury risk and to investigate different possible compensation strategies for reduced core stability. A better understanding of how core stability affects running mechanics and potential compensation strategies used for poor core stability may ultimately contribute to the development of more effective and robust running injury prevention and rehabilitation regimens.

Figure 1.5. Dissertation framework and outline.

1.5 Outline of Upcoming Chapters

The approach for this dissertation is shown in Figure 1.5. Chapter 2 utilizes a novel core stability knockdown protocol (CSKP) developed to experimentally determine the direct downstream effects of reduced core stability on lower extremity joint angles and external joint moments during running, allowing for inferences to be made about how core stability may affect running injury risk. Chapter 3 describes the development and validation of a more physiologically accurate OpenSim full-body model that allows for the creation of dynamic simulations to investigate of full-body dynamics and contributions of the trunk muscles to dynamic tasks. In Chapters 4 and 5, the in vivo experimental kinematic and kinetic data from Chapter 2 is then used as input into 20 musculoskeletal modeling software OpenSim with the model developed in Chapter 3 to create subject-specific simulations of each participant jogging in order to isolate and investigate the consequences of utilizing different possible compensation strategies for reduced core stability. Chapter 4 focuses on the effect of altering running kinematics in response to reduced core stability (kinematic compensation strategy). We examine the effect of a kinematic compensation strategy adopted in response to reduced core stability on internal hip, knee, and lumbar spine joint loads, as well as total energy consumption during running. Chapter 5 investigates the isolated effect of adopting a neuromuscular compensation strategy in response to core muscle fatigue, where one must alter only muscle activation patterns in order to compensate and maintain their initial kinematic state. This chapter examines the effect of a core muscle fatigue and the subsequently adopted neuromuscular compensations strategy on internal hip, knee, and lumbar spine joint loads, as well as total energy consumption during running. Lastly, Chapter 6 outlines the contributions of this dissertation and identifies future research.

Chapter 2 : An Experimental Investigation of the Direct Downstream Effects of Reduced Core Stability on Running Mechanics in Novice Runners

2.1 Abstract

The annual rate of running-related injuries, including patellofemoral pain syndrome

(PFPS), is reported to be as high as 74% and novice runners may be at the highest risk of developing these injuries. Previous research has shown core stability may affect lower extremity function, which has led to the popular notion that insufficient core stabilization may lead to movements that are less efficient and ultimately contribute to injury.

However, the role that core stability plays during running and its influence on injury risk in runners is not well understood. The purpose of this study was to investigate the effect of core stability on running mechanics that have previously been associated with running injuries. Three-dimensional running kinematics and kinetics were collected on 25 healthy, novice runners before and after they performed a core stability knockdown protocol (CSKP), designed to temporarily reduce participants’ core stability in a single testing session. In the participants (N=10) who experienced a meaningful reduction in core stability following the CSKP (≥10% from baseline level), linear mixed models demonstrated that this reduction was significantly associated with an increase in the external peak knee flexion moment (0.81 %BW*h, p=0.001) during the stance phase of

22 running. This biomechanical change has previously been associated with increased patellofemoral contact pressure during running. These results suggest that insufficient core stability in novice runners may be a risk factor for developing PFPS.

2.2 Introduction

Running is an extremely popular physical activity despite the high annual rate of running-related injuries (Marti, Vader et al. 1988; Macera, Powell et al. 1989; van

Mechelen 1992; Buist, Bredeweg et al. 2010; Daoud, Geissler et al. 2012). Novice runners may be the most susceptible to sustaining these injuries, possibly due to a lack of training experience ( Buist, Bredeweg et al. 2010; Tonoli, Cumps et al. 2010; Verhagen

2012; Schmitz, Russo et al. 2014). Scientific literature on the prevention of running- related injuries is scarce, and prior intervention studies focused primarily on modifying training volume, stretching, and the use of external braces or insoles have been unsuccessful at lowering the running injury rate (Buist, Bredeweg et al. 2008; Yeung,

Yeung et al. 2011; Verhagen 2012).

Previous research has identified lower extremity (LE) biomechanical factors that are associated with some common running-related injuries, such as patellofemoral pain syndrome (PFPS), iliotibial band syndrome (ITBS), and tibial stress fractures (TSF).

While these factors may be potential injury mechanisms, very few intervention studies exist with the goal of modifying these parameters to lower injury risk. Some evidence supports gait retraining as a beneficial biomechanical intervention paradigm (Noehren,

Scholz et al. 2010; Crowell and Davis 2011), however this intervention method may not be readily available to all runners. Additional research that addresses the fundamental

23 mechanisms of running-related injuries is therefore needed to design improved treatment and prevention protocols.

Core stability has become an extremely popular topic among clinicians, researchers, and physically active people in general. We define core stability as the body’s ability to control the torso and maintain or resume an equilibrium trunk and pelvis position or state following internal or external perturbations (Zazulak, Hewett et al.

2007). Research has shown that muscle activity of the trunk precedes dynamic movement of the extremities (Hodges and Richardson 1997; Hodges and Richardson 1997; Jamison,

McNally et al. 2013), suggesting the abdominal muscles may contract to provide a stable foundation for production, absorption, and control of force and motion to the extremities.

Consequently, these data have led to the popular theory that insufficient core stability may lead to movements that are less efficient and ultimately musculoskeletal injury

(Fredericson and Moore 2005; Willson, Dougherty et al. 2005; Kibler, Press et al. 2006); however, this theory has not been supported with sufficient scientific evidence.

A limited number of scientific studies have examined the relationship between core stability and running injuries, with the majority being cross-sectional or retrospective designs (Fredericson, Cookingham et al. 2000; Ireland, Willson et al. 2003;

Willson, Kernozek et al. 2011; Shirazi, Moghaddam et al. 2014). The role that core stability plays during running and its influence on running injury risk is accordingly not well understood. Studies from which causal conclusions can be drawn are needed to further understand this relationship and the influence of core stability on injury risk.

The purpose of this study was to investigate the effect of core stability on running mechanics previously associated with injuries in novice runners. Specifically, decreasing core stability in individual runners was hypothesized to result in altered running mechanics previously associated with increased injury risk.

2.3 Methods

2.3.1 Participants

Twenty-five novice runners (23.6±6.8 years; 13 female; 1.74±0.08 m; 69.4±12.6 kg) participated in the study after providing IRB-approved informed consent. Novice runners were defined as running <10 miles per week on average and not playing running sports (e.g., soccer) more than once a week. Other exclusion criteria for this study were as follows: BMI>30; history of lower back pain; musculoskeletal injury within the past 3 months.

2.3.2 Experimental Procedures

Core Stability Knockdown and Running Biomechanics

Three-dimensional overground jogging kinematics and kinetics were collected before and after participants performed a novel core stability knockdown protocol

(CSKP) we previously developed to reduce a person’s core stability temporarily in a single testing session ( Raabe, Monfort et al. 2015). The CSKP consisted of 4 dynamic and 4 isometric exercises that were chosen by our research staff to target both the superficial and deep core musculature with minimal involvement from the lower extremity muscles (Table 2.1). Participants completed all exercises consecutively, each to voluntary exhaustion or until proper form could not be maintained.

Exercise Notes 1-2. Side Bend Repetitions on BOSU® Ball (R & L) Performed on right side followed by the left side. Participants were instructed to keep hips stacked on top of each other and arms crossed. Participants were encouraged to move slowly and controlled through their full range of motion.

3. Back Extension Repetitions on BOSU® ball Participants had arms crossed over the chest and were instructed to relax their legs and gluteal muscles as much as possible. Participants were encouraged to move slowly and controlled through their full range of motion.

4. Swiss Ball Crunches

Participants were instructed to cross arms over the chest and move through a full range of motion where the core muscles were continuously engaged. Participants were encouraged to move slowly and controlled.

5-6. Side Plank Hold (R & L) Performed on right side followed by the left side. Participants were instructed to keep hips stacked on top of each other as high off the ground as possible.

7. Back Extension Hold Participants could place arms off the ground wherever was most comfortable and were instructed to raise their chest off the ground as high as possible, while leaving legs on the ground. Participants were encouraged to fully relax their legs and gluteal muscles.

8. Traditional Plank Hold Participants were instructed to place forearms on the ground and raise their body up so their body was straight and hips were in line with the shoulders.

Table 2.1.Core Stability Knockdown Protocol (CSKP) designed to fatigue both the superficial and deep core musculature. The four dynamic and four isometric exercises were performed in the following order, each to voluntary exhaustion or until proper form could no not be maintained. All exercises were completed consecutively with minimal to no rest between each.

Marker data were collected at 300Hz using 9 Vicon MX-F40 cameras (Vicon

Motion Systems; Oxford, UK) and filtered using a 4th order Butterworth filter at 15Hz.

Ground reaction forces (GRF) were recorded at 1500Hz from Bertec 4060-10 force plates

(Bertec Corp; Columbus, OH, USA). The speed of each trial was monitored using timing gates (Fusion Sport; Sumner Park, QLD, Australia) to ensure all jogging trials were within ±10% of a self-selected, comfortable jogging speed. Retro-reflective markers were placed on the upper and lower body using a modified point cluster technique (Andriacchi,

Alexander et al. 1998; Jamison, Pan et al. 2012) (Fig 2.1). All moments presented are external joint moments. Ten overground jogging trials were collected for each leg, both pre- and post-CSKP. All analyses were performed on stance phase data from both the dominant and non-dominant leg for each participant.

Figure 2.1. Experimental modified point-cluster set used for movement analysis. The lower body Point-Cluster marker set is combined with the upper body Plug-In Gait marker set. The use of a redundant marker set with the point-cluster algorithm reduces the influence of soft tissue artifact (Andriacchi, Alexander et al. 1998).

Primary biomechanical running variables of interest investigated were those previously identified to be associated with running injury risk and included the following: vertical GRF impact peak and average loading rate (VALR) (Davis, Bowser et al. 2015); peak knee flexion (pKF) moment (James 1995); peak knee abduction (pKAbd) angle

(Mizuno, Kumagai et al. 2001; Powers 2003; Chen and Powers 2014); peak knee adduction (pKAdd) moment and impulse (James 1995; Stefanyshyn, Stergiou et al.

2006); peak knee internal rotation (pKIR) angle (Noehren, Davis et al. 2007; Ferber,

Noehren et al. 2010); peak hip adduction (pHAdd) angle (Noehren, Davis et al. 2007;

Pohl, Mullineaux et al. 2008; Ferber, Noehren et al. 2010; Noehren, Hamill et al. 2013),

28 moment and impulse (MacMahon, Chaudhari et al. 2000); and peak hip internal rotation

(pHIR) angle (Lee, Anzel et al. 1994; Powers 2003). Wireless surface electromyography

(EMG) (Telemyo DTS; Noraxon USA, Inc; Scottsdale, AZ) was measured during the

CSKP using disposable surface electrodes. EMG was placed bilaterally on the external obliques (EO) and internal obliques (IO), and unilaterally on the dominant-side rectus abdominis (RA), and L5 lumbar extensor (L5). Dual electrodes were used for the EO, RA and L5 muscles and two single electrode discs were used for the IO muscles (Jamison,

McNally et al. 2013). Pre-gelled (Ag/AgCl), surface electrodes (A10011/A10005;

Vermed, Inc; Bellows Falls, VT, USA) were placed as recommended by McGill to best reflect deep muscle activity (McGill, Juker et al. 1996) or directly on the most prominent aspect of the muscle belly and oriented parallel to the muscle fibers. Electrode locations were shaved, if necessary, and cleaned with alcohol pads. The median frequency (MedF) of the raw EMG signal for all core muscles was analyzed during the isometric exercises of the CSKP in order to obtain a measurement of core muscle fatigue induced by the

CSKP. A decrease in the MedF of an EMG signal has been directly related to the level of muscle fatigue (De Luca 1983).

Core Stability Assessment

A novel unstable quiet sitting test (QST) developed by our research team was used to quickly and objectively quantify each participant’s core stability in all planes of motion in a way that could also be replicated in a clinical setting (Fig. 2.2). The QST is a modification of the quiet standing task commonly used to measure postural stability

(Madigan, Davidson et al. 2006), in order to isolate control of the lumbar spine and trunk

29 from adjustments in the lower extremity joints. Postural control of the trunk has previously been investigated during sitting on an unstable chair with a rigid hemisphere beneath the seat (Cholewicki, Polzhofer et al. 2000; van der Burg, van Wegen et al. 2006; van Dieen, Luger et al. 2012; Hendershot and Nussbaum 2013), however this chair is complex and may not be easily replicable by other researchers or clinicians. In the QST, participants sat on top of the rounded end of a BOSU® ball positioned on a platform high enough so that their feet did not touch the ground. This platform was placed on top of two

Bertec 4060-10 force platforms and center of pressure (CoP) data were recorded.

Participants were instructed to sit as still as possible for 60 seconds with their eyes closed and arms crossed while performing a secondary task of counting backwards by 4’s. This secondary task was used to provide a more functional measure of core stability, as real- life situations almost always require attention to be divided between multiple tasks

(Moghadam, Ashayeri et al. 2011). The starting number was incremented by one for each trial.

Figure 2.2. Unstable quiet sitting task (QST) used to quantify core stability. Participants are instructed to sit as still as possible for 60 seconds while simultaneously counting down by 4’s. The chair is positioned on top of force platforms and center of pressure excursion is recorded during the task.

Core stability was quantified during the QST through measurement of the total

CoP path length, also known as total CoP excursion (CoPexc) during the sitting task.

CoPexc is a common parameter used to quantify postural control (Cholewicki, Polzhofer et al. 2000; Radebold, Cholewicki et al. 2001; Lafond, Corriveau et al. 2004; van der

Burg, van Wegen et al. 2006; Fisher 2010). Each participant performed one practice trial 31 of the QST and then performed five trials before the CSKP and only three trials after to minimize recovery. The median CoPexc was used for analysis. The QST has been shown to have very good intra- and inter-rater reliability (ICC(2,1): 0.877 and 0.813, respectively) (Raabe, Teater et al. 2016). We defined an increase in CoPexc ≥10% from baseline (pre-CSKP) as a meaningful decrease in core stability. This threshold was chosen to represent a moderate decrease in postural stability, as impairments in postural control of ≥25% have been reported in populations with severe balance deficits (elderly adults (Maki, Holliday et al. 1990), patients with Parkinson’s Disease and a history of falls (van der Burg, van Wegen et al. 2006)) when compared to matched healthy controls.

2.3.3 Statistics

A sample size of 25 was calculated a priori to provide at least 80% power to detect an effect size of 0.75 standard deviation change based on a two-sided paired t-test at a significance level of 0.01 (Bonferroni adjustment for up to 5 primary endpoints)

(Lenth 2006), which is an appropriate and clinically-relevant effect size for the biomechanical variables of interest based on previous data (MacMahon, Chaudhari et al.

2000; Stefanyshyn, Stergiou et al. 2006; Noehren, Davis et al. 2007).

Statistical analyses were performed in JMP® (JMP® Pro, Version 12.2.0; SAS

Institute Inc., Cary, NC). Paired two-sided t-tests were performed to test for a significant change in core muscle fatigue following the CSKP. Linear mixed models were used to evaluate the effect of a change in core stability state (pre- or post-CSKP) on the primary variables of interest. Core stability state was treated as a fixed effect; participant and leg were treated as random effects. The use of a random subject effect and a random leg

32 effect allows for the examination of the effect of a change in core stability on running parameters within an individual and within legs. The best model (based on AICc and BIC criterion (Burnham and Anderson 2004)) for each of the running parameters was used for analysis. The significance level for all tests was set at α=0.05 and corrections for multiple comparisons were not made due to the exploratory nature of this study. Mixed model analyses were restricted to participants whose core stability meaningfully decreased

(CoPexc≥10%) post-CSKP.

2.4 Results

2.4.1 Muscle Fatigue and Core Stability

Ten of the 25 participants (21.5±2.7 years; 5 female; 1.73±0.09 m; 68.1±11.9 kg) exhibited a meaningful decrease in core stability post-CSKP. Figure 2.3 shows the distribution of CoPexc and MedF data for the group that experienced a reduction in core stability post-CSKP (N=10) and the group that did not (N=15). A positive change in

CoPexc corresponds to reduced stability and a negative change in MedF corresponds to more muscle fatigue. Figure 2.3 also includes the results for two-sided paired t-tests used to test for significant changes in core stability (COPexc) and average core muscle fatigue

(MedF) post-CSKP. The group of participants that experienced a meaningful decrease in core stability exhibited a significant decrease in the average core MedF (p<0.0001) and a significant increase in COPexc (p<0.0001), indicating this group had both significantly less core stability and significantly more core muscle fatigue post-CSKP. The average percent change in MedF for the group was -17.5%, indicating that the CSKP elicited a moderate level of muscle fatigue.(Hart, Fritz et al. 2006; Szucs, Navalgund et al. 2009)

The remaining group of participants had no significant change in COPexc (p=0.125) and a significant decrease in MedF (p=0.0001) post-CSKP, indicating this group was significantly fatigued but experienced no change in core stability. This group was also moderately fatigued with an average percent change in MedF of -14.7%.

Figure 2.3. Box plots showing the distribution of center of pressure excursion (CoPexc) and median frequency (MedF) data for the group that experienced a meaningful reduction in core stability post-CSKP (N=10) and the group that did not (N=15). Data is expressed as the percent change in the parameter from pre- to post-CSKP. The p-value corresponds to a two-sided paired t-test used to determine if the parameters investigated were significantly different after the CSKP. A positive change in CoPexc corresponds to reduced stability and a negative change in MedF corresponds to more muscle fatigue.

2.4.2 Biomechanical Running Parameters

The mean changes in biomechanical running parameters from pre- to post-CSKP and the mixed model effects are shown in Table 2.2 along with corresponding p-values.

Since the best model for each of the running parameters was used for analysis, an interaction effect between CSKP and leg was included for some parameters. Table 2.2 includes data from both legs for all participants in only the group that experienced a meaningful decrease in core stability.

A significant positive effect of the CSKP was found on peak knee flexion (pKF) moment during running (6% increase, p=0.001), indicating running in a state of reduced core stability (post-CSKP) resulted in an increased pKF moment during stance.

Additionally, the CSKP was found to have a significant negative effect on peak knee abduction (pKAbd) angle (defined as the negative knee frontal plane angle) and the knee adduction (KAdd) impulse, meaning that the CSKP resulted in an increased pKAbd angle (43% increase, p=0.005) and decreased KAdd impulse (5% decrease, p=0.046).

There was no evidence of an effect of the CSKP on any of the other running parameters investigated.

For exploratory analysis, mixed models were also created for each running parameter separately for each leg. Table A.1 shows the mixed-model results for the dominant leg, non-dominant leg, and combined legs separately. It is important to consider the reduction in sa mple size with individual legs.

For further analysis, in order to attempt to separate the effect of core fatigue from core stability, the parameters that were found to be significantly affected by the CSKP in

35 the subset of participants that had reduced core stability (CS group) were also investigated in the non-core stability (NCS) group, the group that was fatigued but did not have a meaningful reduction in stability following the CSKP. Table 2.3 shows the changes in these parameters compared between groups and the p-values corresponding to the CSKP effect from linear mixed model analysis. In the NCS group (N=15), there was a significant effect of the CSKP on the pKAbd angle (64% increase, p=0.0002) and no effect of the CSKP on either the pKF moment or KAdd impulse.

Table 2.2. Initial values (pre-CSKP) and mean changes from pre- to post-CSKP and for each of the biomechanical running parameters for the subset of participants that experienced a meaningful reduction in core stability following the CSKP (CS group, N=10). The best linear mixed model for some of the running parameters also included an interaction effect between CSKP and leg this effect is also included if applicable. The CSKP effect characterizes the relationship between a reduction in core stability and the respective biomechanical variable. A reduction in core stability was associated with an increased peak knee flexion moment, an increased peak knee abduction angle (negative frontal plane angle), and a decreased knee adduction impulse during running. Vertical Knee Knee Knee Knee Hip Hip Hip Average Knee Hip Flexion Adduction Adduction Internal Adduction Adduction Internal Loading Abduction Adduction Moment Moment Impulse Rotation Moment Impulse Rotation Rate Angle [°] Angle [°] [%BW*h] [%BW*h] [%BW*h*s] Angle [°] [%BW*h] [%BW*h*s] Angle [°] [BW/s]

Pre-CSKP 11.81 ± 13.45 ± -2.84 ± 6.93 ± 0.82 ± 10.19 ± 12.74 ± 10.09 ± 1.42 ± 14.67 ± Value

37 2.54 2.54 2.65 1.94 0.30 2.94 4.59 2.85 0.51 9.27 (mean±SD)

Mean -0.59 ± 0.81 ± -1.22 ± -2.04 ± -0.05 ± 0.85 ± 0.10 ± -0.04 ± -2.04 ± 0.01 ± 0.13 Change 0.40 0.20 0.39 0.10 0.02 0.58 0.54 0.02 0.99 (0.94) (p-value) (0.16) (0.001) (0.005) (0.19) (0.046) (0.16) (0.85) (0.07) (0.053)

CSKP and -0.41 ± 0 10 ± 0.02 ± 0.02 ± Leg 0.19 0.10 0.01 0.01 Interaction (0.049) (0.08) (0.028) (0.04) Effect Results are expressed as the mean ± standard error (p-value), unless otherwise stated. The p-value indicates the significance of the CSKP model effect. Significant effects (p<0.05) are bolded. All GRF loading variables were normalized to participant body weight (BW) and moments and impulses were normalized to BW and height (h). Peaks angles and moments were analyzed. Table 2.3. Mean change ± standard error (SE) from pre- to post-CSKP and the p-values corresponding to the CSKP mixed-model effect for the biomechanical variables found to be significant in the reduced core stability (CS) group compared to the changes seen in the non-core stability (NCS) group.

Change ± SE p-value

Peak Knee Flexion Moment [%BW*h] CS 0.81 ± 0.20 0.001 NCS 0.30 ± 0.20 0.166

Peak Knee Abduction Angle [°] CS -1.22 ± 0.39 0.005 NCS -1.11 ± 0.26 0.0002

Knee Adduction Impulse [%BW*h*s] CS -0.05 ± 0.02 0.046 NCS -0.03 ± 0.02 0.245 Significant effects (p<0.05) are bolded. All GRF loading variables were normalized to participant body weight (BW) and moments and impulses were normalized to BW and height (h). Peaks angles and moments were analyzed.

2.5 Discussion

2.5.1 Implications on Injury Risk in Novice Runners

The results of this study supported our hypothesis that a decrease in core stability would result in altered running mechanics previously associated with increased running injury risk (Fig. 2.4). The CSKP had significant positive effect on the pKF moment during running in the CS group, suggesting that reduced core stability may lead to an increase in this variable. As the external knee flexion moment during stance must be balanced by the quadriceps, an increase in this moment results in a compensatory increase in the quadriceps workload (Andriacchi, Kramer et al. 1985; James 1995). Since

38 the quadriceps tendon attaches directly to the patella, an increased workload of the extensor mechanism directly increases the load placed on the patellofemoral joint, which may consequently lead to PFPS (James 1995). This effect was not seen in the NCS group who was fatigued but had no meaningful change in core stability following the CSKP, further supporting that a reduction in core stability may be driving this change in running mechanics.

The CSKP also had a significant negative effect on pKAbd angle during stance.

Since abduction is defined as negative in the frontal plane, this corresponds to an increase in the pKAbd angle following the CSKP. Increased knee abduction is associated with an increased dynamic Q-angle (Powers 2003), which has been shown to result in patellar maltracking, increased lateral patellar contact forces (Mizuno, Kumagai et al. 2001) and consequently, increased risk of PFPS (Messier, Davis et al. 1991; Mizuno, Kumagai et al.

2001; Powers 2003; Chen and Powers 2014). However, this effect was significant not only in the CS group that experienced a meaningful reduction in core stability following the CSKP, but also in the NCS group that did not experience a change in core stability.

These results suggest that this effect may be driven from a source other than a change in core stability. Muscle activity of the gluteus medius muscle was also measured during the

CSKP and upon further analysis, it was found that there was no significant change in the

MedF of this muscle post-CSKP in either participant group, indicating that this muscle was not significantly fatigued following the CSKP. Therefore, these data suggest that the significant increase in pKAbd angle following the CSKP is likely driven by some factor

39 other than reduced core stability or hip abductor muscle fatigue, such as cognitive fatigue.

Figure 2.4. Schematic of study results and their expected effect on patellofemoral contact mechanics. An increased knee flexion moment (Mkf, bottom left) must be balanced by an increase in the quadriceps workload, which increases the quadriceps and patellar ligament tensile force (Fq, Fplig) acting on the patella and results in an increased compressive patellofemoral joint force (FPFJ). Increased knee abduction (bottom right) during running alters the line of action of the Fq and Fplig potentially resulting patellar maltracking and an increased lateral component of the FPFJ. These changes in patellofemoral contact mechanics may increase risk of developing patellofemoral pain in novice runners. * indicates a significant difference between pre- and post-CSKP conditions.

The one finding that did not support our hypothesis was that the CSKP was associated with a decreased KAdd impulse. A decrease in the KAdd impulse is indicative of reduced cumulative frontal plane loading of the knee, possibly reducing PFPS injury risk (Stefanyshyn, Stergiou et al. 2006). This effect was only significant in the participant group that experienced a meaningful reduction in core stability. It is possible that participants were adapting their gait to limit frontal plane knee loading during running with reduced core stability, though future study is necessary to examine this possibility.

It has yet to be established the specific magnitude of increased pKF moment and

KAbd angle during running that may elicit PFPS. There is in-vitro evidence that increased repetitive stress placed on the patella accelerates retropatellar cartilage damage

(Zimmerman, Smith et al. 1988). Heino Brechter and Powers used a biomechanical model to compare patellofemoral joint stress during walking between participants with and without PFPS and found the PFPS group had over 2x greater peak stress than the control group (3.13 MPa vs 6.61 MPa) (Heino Brechter and Powers 2002). Since patellofemoral joint stress is difficult to measure in vivo, the external knee flexion moment is commonly used as a surrogate measure (Willson and Davis 2008).

Additionally, an increase in the knee abduction angle is associate with an increased dynamic Q-angle (Powers 2003) and cadaveric research has shown that a 10° increase in the Q-angle results in a 45% increase in peak patellar pressure at 20° of knee flexion

(Huberti and Hayes 1984).

Two of the three biomechanical running parameters that were observed to be significantly affected by the CSKP have been previously identified to increase PFPS risk.

Out of these parameters, the one that appears to be primarily driven by a reduction in core stability is an increase in the pKF moment. The majority of previous research studies linking core stability to PFPS risk involve hip muscle dysfunction as the driving risk factor (Ireland, Willson et al. 2003; Barton, Lack et al. 2013; Shirazi, Moghaddam et al.

2014). One cross-sectional study has shown trunk side flexion strength was significantly less in symptomatic PFPS patients when compared to healthy controls (Cowan, Crossley et al. 2009). The present investigation is the first experimental study, to the authors’ knowledge, to create a quantified deficit in core stability and identify associated changes in biomechanical parameters during running that may increase the risk for developing

PFPS.

2.5.2 Limitations

There are limitations that must be considered when interpreting these results. The

CSKP only resulted in a meaningful drop in core stability for a portion of our participants, which may have been partly due to the reliance on volitional exhaustion during the CSKP. This reliance may have resulted in greater-than-desired variability in the force production of the targeted core muscles or potential confounding by cognitive fatigue. An increased cognitive load has been associated with increased knee abduction during a landing task (Méjane, Faubert et al. 2016), so cognitive fatigue induced by exhaustive physical exercise may be driving the significant increase in kAbd angle found in both the CS and NCS group. Additionally, core muscle fatigue may not directly result

42 in a loss of stability for all individuals. All participants (N=25) experienced at least some level of core muscle fatigue following the CSKP, but only a select subset (N=10) also experienced a loss of core stability, suggesting that an individual’s response to muscle fatigue may depend on many factors (Raabe, Monfort et al. 2016). Previous research examining the effect of shoulder muscle fatigue on upper extremity movement found that it was possible for participants to maintain movement stability in the presence of muscle fatigue (Gates and Dingwell 2011). Additionally, the standard limitations of skin-based passive marker motion capture systems apply, though the use of a redundant marker set with the point-cluster algorithm to reduce the influence of soft tissue artifact (Andriacchi,

Alexander et al. 1998) and a within-subject research design should have minimized the effects of these limitations.

2.6 Conclusion

It was observed that a protocol resulting in a temporary decrease in core stability was associated with an increased peak knee flexion moment (pKF) during the stance phase of running (Fig. 2.4). Therefore, insufficient core stability in novice runners may be a risk factor for PFPS. Further studies are needed to determine whether interventions aimed at improving core stability can alter the biomechanical variables in this study known to increase risk for PFPS.

2.7 Acknowledgements

This work was supported by funding from NIAMS R03AR065215 and NSF

GRFP DGE-1343012. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. 43

Chapter 3 : An Investigation of Jogging Biomechanics using the Full-Body Lumbar Spine Model: Model Development and Validation

3.1 Abstract

The ability of a biomechanical simulation to produce results that can translate to real-life situations is largely dependent on the physiological accuracy of the musculoskeletal model. There are a limited number of freely-available, full-body models that exist in

OpenSim, and those that do exist are very limited in terms of trunk musculature and degrees of freedom in the spine. Properly modeling the motion and musculature of the trunk is necessary to most accurately estimate lower extremity and spinal loading. The objective of this study was to develop and validate a more physiologically accurate

OpenSim full-body model. By building upon three previously developed OpenSim models, the Full-Body Lumbar Spine (FBLS) model, comprised of 21 segments, 29 degrees-of-freedom, and 324 musculotendon actuators, was developed. The five lumbar vertebrae were modeled as individual bodies, and coupled constraints were implemented to describe the net motion of the spine. The eight major muscle groups of the lumbar spine were modeled (rectus abdominis, external and internal obliques, erector spinae, multifidus, quadratus lumborum, psoas major, and latissimus dorsi), and many of these muscle groups were modeled as multiple fascicles allowing the large muscles to act in multiple directions. The resulting FBLS model’s trunk muscle geometry, maximal

44 isometric joint moments, and simulated muscle activations compared well to experimental data. The FBLS model was made freely available

(https://simtk.org/home/fullbodylumbar) for others to perform additional analyses and develop simulations investigating full-body dynamics and contributions of the trunk muscles to dynamic tasks.

3.2 Introduction

Dynamic simulations of human movement are a beneficial addition to experimental data, as they enable researchers to conduct biomechanical investigations involving parameters of the neuromusculoskeletal system that are difficult or impossible to examine using experiments alone. However, the ability of a simulation’s results to translate to real-life situations is dependent on the physiological accuracy of the musculoskeletal model.

OpenSim is a freely available, open-source musculoskeletal modeling software

(Delp, Anderson et al. 2007) that allows users to develop and analyze dynamic simulations of human movement. The open-source nature of the software results in a large database of previously built, validated, and tested musculoskeletal models for other users to expand upon. Despite this large database, very few full-body models exist in

OpenSim, and those that do exist are very limited in terms of trunk musculature and degrees of freedom in the spine (Hamner, Seth et al. 2010; Caruthers, Thompson et al.

2013).

Properly modeling trunk motion and muscle activations is necessary to most accurately estimate lower extremity loads as well as spinal loading. Additionally, a

45 physiologically relevant trunk is necessary to investigate the role of core strength and stability in dynamic movements, a topic that has received increasing clinical and scientific interest over the past decade (Willson, Dougherty et al. 2005; Kibler, Press et al. 2006; McGill 2010; Chaudhari, Jamison et al. 2012; Jamison, McNally et al. 2013;

Ferber, Bolgla et al. 2015). The purpose of this study was to develop and validate a more physiologically accurate OpenSim full-body model with extensive trunk musculature and degrees of freedom in the lumbar spine.

3.3 Methods

3.3.1 Model Development

The Full-Body Lumbar Spine (FBLS) model (Fig. 3.1) was developed by combining three previously built OpenSim models: Hamner’s full-body model (Hamner,

Seth et al. 2010) for the base model, Christophy’s lumbar spine model (Christophy, Faruk

Senan et al. 2012) for the torso, and Arnold’s model (Arnold, Ward et al. 2010) of the patella. Brief details about model development will be presented here and more details can be found in Appendix B (Section B.1.1).

Figure 3.1. The full-body lumbar spine (FBLS) model. The model consists of 21 segments, 29 degrees-of-freedom, 324 musculotendon actuators, and wrapping surfaces (blue).

When combining models together extreme care must be taken to ensure all models are scaled to the same size person, so that when they are combined all mass and inertial properties are consistent across the models. After scaling the component models, discrepancies still remained since Hamner’s model and Christophy’s model used different ribcage and pelvis geometry. Consequently, all attachment points of the 224 trunk muscle fascicles to the ribcage and pelvis had to be adjusted to their appropriate physiological

47 locations in the FBLS model. It was important that these were attached with close inspection to ensure the most anatomically correct muscle paths, because incorrect attachments and paths would result in non-physiological estimations of muscle activations and forces.

The resulting FBLS model is comprised of 21 segments, 29 degrees-of-freedom, and 324 musculotendon actuators. The five lumbar vertebrae are modeled as individual bodies, each connected by a 6 degree-of-freedom joint (Christophy, Faruk Senan et al.

2012). After 27 coupling constraints are imposed, the net lumbar movement is described as three rotational degrees-of-freedom: flexion-extension, lateral bending, and axial rotation (Christophy, Faruk Senan et al. 2012). The rigid torso (lumped thoracic and cervical vertebrae, ribcage, scapulae, and head) is connected to the first lumbar vertebrae by a one degree-of-freedom rotational joint that allows for additional axial rotation of the torso, if necessary. While some detailed musculoskeletal models of the spine exist (de

Zee, Hansen et al. 2007; Christophy, Faruk Senan et al. 2012; Bruno, Bouxsein et al.

2015), the FBLS model is the first full-body OpenSim model to describe the trunk musculature in this level of detail. The eight major muscle groups of the lumbar spine that are modeled include the erector spinae (ES), rectus abdominis (RA), external obliques (EO), internal obliques (IO), multifidus (MF), quadratus lumborum (QL), psoas major (PS), and latissimus dorsi (LD). Every muscle group is modeled as multiple fascicles with different lines of action to account for the fact that most of the trunk muscles are large and can act in multiple directions (Christophy, Faruk Senan et al.

2012). The ES is defined as the iliocostalis lumborum (IL) and the longissimus thoracis

(LT), each of which have a rib/thoracic (IL_R, LTpT) and lumbar component (IL_L,

LTpL). Wrapping surfaces are also included in the model to ensure physiological muscle lines of action.

3.3.2 Model Validation

The model validation process includes comparing model parameters and simulations to experimental data to ensure that they represent the physical phenomena of interest (Hicks, Uchida et al. 2015). We validated the FBLS model through the three phases described in the following sections that have previously been used to validate the capability of a musculoskeletal model to produce a dynamic simulation (Holzbaur,

Murray et al. 2005; Arnold, Ward et al. 2010; Hicks, Uchida et al. 2015).

Validating model parameters

The first step of the validation process is to validate model parameters by comparing them to experimental values measured in vivo or in cadavers. Authors of the three individual models have each done extensive literature searches to determine the individual segment and muscle properties of their models and very few of these were altered when developing the FBLS model. The only individual muscle parameter altered in development of this model was the maximum isometric force property in several of the trunk and lower extremity muscles that were found to be too weak for simulations of jogging (listed in Appendix B, Table B.1). As mentioned previously, since several of the trunk muscle attachment points were altered it was necessary to validate the trunk muscle geometry to ensure that muscle attachments and lines of action were physiologically relevant. To do this, we compared the trunk muscle sagittal plane moment arms at zero

49 degrees trunk flexion to those measured experimentally in the literature (Jorgensen,

Marras et al. 2001).

Validating muscle function

Next, we validated muscle function by examining the model’s moment generating capacity about a given joint to ensure the moment is comparable to experimental results

An individual muscle’s contribution to the total moment generating capacity about a joint, also known as a maximum isometric joint moment, is calculated in OpenSim as the product of its moment arm and maximum isometric force at a given joint position. The model’s total maximum isometric moment for a given joint is the sum of the individual moments produced by each muscle that can contribute to the moment, assuming maximal activation, over a range of joint angles. Since minimal changes were made to the lower extremities during development of this model, validation was focused on function of the trunk musculature about the L5-S1 joint.

Maximal isometric trunk joint moments measured experimentally in our laboratory and similar data reported in the literature were compared to the moment generating capacity of our model. Seven healthy adult males (mass = 79.30 ± 9.18 kg, height = 1.79 ± 0.07 m, age = 22.43 ± 2.89 y) participated after providing IRB-approved consent. To experimentally measure maximum isometric joint moments for trunk flexion, extension, axial rotation, and lateral bending, a custom device (Jamison, McNeilan et al.

2012) was used to fix a participant’s pelvis while in a semikneeling position. While in this position, an instrumented cable was attached to an upper body harness positioned on the participant at approximately the level of the 10th thoracic vertebra. The length of the

50 cable was adjusted to position the participant at a specified trunk angle. Participants were instructed to pull maximally against the cable for five seconds and the maximum force generated during the task was recorded and used to calculate the maximal isometric joint moment. One practice trial was performed followed by one maximal exertion. To experimentally measure axial rotation strength we positioned participants in the semikneeling position on top of force plates embedded into the floor. Their upper body was fixed against a rigid bar at approximately an inch below the clavicle to prevent any upper body movement. Similarly, they were instructed to maximally twist against this rigid bar and the free moment of the ground reaction force was recorded. The maximum value for this moment was considered the experimental maximum isometric axial rotation moment.

Validating simulations

Lastly, model simulations were validated by comparing model muscle activations estimated during Static Optimization to experimentally measured surface electromyography ( EMG) during overground jogging at a comfortable speed (2.48 ± 0.25 m/s). Kinematics, kinetics, and EMG during jogging were collected for one healthy participant (male, mass = 100.93 kg, height = 1.85 m, age = 30 y) as part of another IRB- approved study. The generic musculoskeletal model was scaled to match the anthropometry of the study participant. Inverse Kinematics, Inverse Dynamics (using the residual reduction algorithm), and Static Optimization (SO) were performed in OpenSim

(Delp, Anderson et al. 2007) to estimate individual muscle activations during jogging

(Fig. 1.4). Surface electromyography (EMG) was performed on the following muscles as

51 described by McGill et al. (McGill, Juker et al. 1996) or directly on the muscle belly unilaterally on the dominant side (bilaterally for obliques): RA, EO, IO, erector spinae

(L5), gluteus maximus (GMax), gluteus medius (GMed), rectus femoris (RF), vastus lateralis (VL), vastus medialis (VM), biceps femoris (BF), semitendinosus (ST), lateral gastrocnemius (LG), soleus (Sol), and tibialis anterior (TA). Before analysis, EMG was processed (10-500Hz band-pass filtered, rectified, and RMS smoothed with 60 ms window) and normalized to the peak activation over the gait cycle. A 40 ms delay was applied to processed EMG to account for electromechanical delay between surface EMG and force production (Arnold, Hamner et al. 2013). Normalized EMG was compared to simulated muscle activations, which are defined between 0 and 1. For muscle groups that are modeled as multiple fascicles, we compared the average activation of all the fascicles in the muscle group to the corresponding EMG.

3.4 Results and Discussion

3.4.1 Model Parameters

Sagittal plane moment arms for the RA, EO, and IO muscle fascicles with respect to each of the lumbar vertebral levels are shown in Table 3.1. The moment arm of each fascicle and the average (AVG) moment arm for a group of muscle fascicles was compared to the moment arm for the respective muscle group recorded by Jorgensen et al. experimentally using magnetic resonance imaging (Jorgensen, Marras et al. 2001). It is important to note that often, one experimentally measured moment arm for an entire muscle group is compared to all of the individual fascicles for that muscle group in the model. The RA, EO, and IO muscle groups were found to have fascicles with a moment

52 arm within one SD of the experimentally collected moment arm for almost all joint levels. The FBLS model was scaled to the average height and weight of the subjects for which experimental data was collected (Jorgensen, Marras et al. 2001), however ribcage geometry and spinal curvature of the model are not subject specific. Additionally, if multiple fascicles in a muscle group cross a given joint level, it is unknown for which of these fascicles experimental data was collected. These limitations may explain why model moment arms compare well to experimental data at certain joint levels but not as well to others. This same analysis was completed for all trunk muscle fascicles (shown in

Appendix B, Table B.2) and similar results were found.

Moment Arm (mm) with Respect to Given Lumbar Vertebral Joint Level Fascicle L1-L2 L2-L3 L3-L4 L4-L5 L5-S1 Rectus Abdominis

RA 73^^ 76^^ 64^^ 64* 75* External Oblique

EO1 74* 85^^ 85^^ 90^^ 40* EO2 70* 80^^ 81^^ 87^^ 37* EO3 55** 67** 70^^ 80^^ 36* EO4 41^^ 54* 61^^ 73^^ 34* EO5 7^^ 10^^ 14* 25* 45* EO6 6^^ 4^^ 2^ 15* 36* AVG 42^^ 50* 52^ 62^^ 38* Internal Oblique

IO1 N/A N/A N/A N/A 37* IO2 N/A N/A N/A N/A 30** IO3 N/A N/A N/A N/A 36* IO4 57^^ 53^ 39* 31* 20^ IO5 28^^ 22^^ 12^^ 8^ 16~ IO6 1^^ 5^^ 10^^ 7^ 6^^ AVG 29^^ 27^^ 20^ 16** 24^ Table 3.1. Sagittal plane moment arms for each of the RA, EO, and IO muscle group fascicles. A *,**, ^ or ^^ signifies the model’s moment arm was within 1 standard deviation (SD), 1.01-1.5 SD, 1.51-2 SD, or 2.01+ SD, respectively, of Jorgensen et al., 2001.

3.4.2 Muscle function

Figure 3.2 shows maximum isometric joint moments for the trunk degrees-of- freedom in the model compared to experimental data collected in this study and data in the literature examining trunk strength at multiple joint angles (Kumar, Dufresne et al.

1995; Khalaf, Parnianpour et al. 1997; Keller and Roy 2002). While the model’s joint moments do not correspond exactly with experimental data, the general behavior is

54 comparable. These differences between the model and experimental data may arise due to slight differences between muscle physiological cross sectional area and tendon slack length parameters in the model and in the population for which experimental data was collected. These values in the model are generally acquired through cadaveric studies, which may not be representative of a young, healthy experimental population. The maximum isometric force properties for several muscles have been scaled up to account for this difference, but this will not account for all discrepancies.

Additionally, when calculating the maximum isometric joint moment in the model, we assumed all muscles that could contribute to the moment are doing so with maximal activation. Experimentally, this may not be the case but it is not possible from this data to discern varying levels of activation to incorporate into the model.

Trunk and lower extremity simulated jogging kinematics are shown in Figure 3.3 and simulated joint moments are shown in Figure 3.4. These joint angles and moments compare well to those measured experimentally in the literature (Novacheck 1998;

Brown, O’Donovan et al. 2014), as well as those previously reported using full-body musculoskeletal simulation models (Hamner, Seth et al. 2010). Figure 3.4 compares trunk and lower extremity (LE) joint moments during jogging simulated by Hamner et al 2010 to the moments simulated using the FBLS model. There is some variability between the data due to comparing data from two different subjects, but overall the simulated moments compare well. Additionally, it is important to note that Hamner’s model had no degrees of freedom in the spine. Figure 3.5 shows the EMG experimentally collected and simulated activations. Generally, the LE muscle activations tend to compare well to EMG measured in this study and EMG reported during jogging in the literature (Cappellini,

Ivanenko et al. 2006), while some of the trunk muscles compare better than others. There may be more variability in trunk muscle activation than LE muscle activation during jogging because the trunk muscles primarily act as stabilizers during running and produce forces considerably lower in magnitude. Additionally, some discrepancies between EMG and simulated activations occur because SO is a frame-by-frame solver so anticipatory activations are not reflected in the simulated activations. If anticipatory actions are of interest, Computed Muscle Control (CMC) should be used instead.

Figure 3.3. Trunk and lower extremity jogging kinematics simulated using the FBLS model over one right foot gait cycle. Abbreviations: right (R) or left (L) anterior superior iliac spine (ASIS), extension (ext), flexion (flex), plantarflexion (plantarflex), dorsiflexion (dorsiflex). The direction of lateral bending and pelvis list is with reference to the stance leg.

Figure 3.4. Trunk and lower extremity (LE) joint moments during jogging simulated by Hamner et al 2010 and using the FBLS model. Note the two studies simulate different subjects with different kinematics. Abbreviations: extension (ext), flexion (flex), plantarflexion (plantarflex), dorsiflexion (dorsiflex). The direction of lateral bending and lumbar rotation is with reference to the stance leg.

Figure 3.5. EMG for all collected muscles (dashed line) compared to model activations estimated in OpenSim (solid line) using Static Optimization. The black vertical line represents toe-off. For simulated activations, the average activation for a muscle group modeled as multiple fascicles is reported, as noted.

3.4.4 Limitations

Limitations exist that should be considered when using this model. With 324 musculotendon actuators, the computational cost to create simulations with this model is higher than simpler models. This increased cost is a trade-off for a more physiologically accurate model. Additionally, spinal curvature in the FBLS model was not subject- specific. Future studies should consider using imaging techniques to accomplish this

(Zhou, McCarthy et al. 2000), as spinal curvature has been shown to affect vertebral load magnitudes (Bruno, Anderson et al. 2012).

The model is not yet suited for CMC or Forward Dynamics. This is likely because the maximum force property of some muscles were altered (described in Appendix B,

Table B.1) while the passive muscle property remained the same. The simulation results presented here are not affected by this discrepancy since the SO algorithm excludes the passive muscle force when calculating instantaneous muscle forces (Hicks and Dembia

m 2014). CMC does not exclude passive muscle forces, so increasing F 0 without also adjusting the passive muscle force property will lead to high passive muscle forces in the model (Thelen and Anderson 2006). SO has been shown to predict muscle forces during walking and running that are comparable to those using CMC (Lin, Dorn et al. 2012), and

SO is considered more robust and computationally efficient than CMC (Mokhtarzadeh,

Perraton et al. 2014). Therefore, SO remains a valuable tool to investigate the behavior of muscle forces and activations during dynamic tasks.

While some trunk EMG was collected in this study, it was limited. EMG was not collected for some of the trunk muscles in the FBLS model (LD, QL, PS, MF) to

61 compare with the estimated activations. Future studies should consider using more surface and fine-wire EMG on the trunk to further validate upper body muscle activations and force estimates in this model.

3.5 Conclusion

An OpenSim full-body musculoskeletal model was developed with detailed trunk musculature and degrees of freedom in the lumbar spine. The FBLS model is the first

OpenSim model to the authors’ knowledge to combine a complex model of the spine, involving detailed trunk musculature and degrees of freedom, with a well-established

OpenSim lower extremity model. Future studies may explore integrating models of the spine that incorporate even higher complexity than the one used in this study (de Zee,

Hansen et al. 2007; Bruno, Bouxsein et al. 2015). The FBLS model was made freely available (https://simtk.org/home/fullbodylumbar) for others to perform additional analyses and develop simulations investigating full-body dynamics and contributions of the trunk muscles to dynamic tasks.

3.6 Acknowledgements

We would like to thank the National Institute for Arthritis and Musculoskeletal and Skin Diseases for providing funding to support this study (Award Number:

R03AR065215).

Chapter 4 : The Biomechanical Effects of Utilizing a Kinematic Compensation Strategy in Response to Reduced Core Stability during Running

4.1 Abstract

It is widely believed that insufficient core stability may lead to less efficient movements and ultimately musculoskeletal injury. It is unknown how runners lacking sufficient core stability compensate for this deficiency and the consequences of chosen compensation strategies. Altering movement patterns may be one strategy commonly used by runners to compensate for poor core stability. The purpose of this study was to identify the biomechanical consequences of altering running kinematics in response to reduced core stability and to determine if the effect of this kinematic compensation strategy is dependent on the presence of core muscle fatigue. Eight novice runners jogged overground in a motion capture laboratory before and after performing a fatiguing core stability knockdown protocol (CSKP). Using this experimentally collected data, subject- specific kinematically-driven jogging simulations were developed for each participant in four different conditions (pre-CSKP kinematics, no core muscle fatigue; pre-CSKP kinematics, core muscle fatigue; post-CSKP kinematics, no core muscle fatigue; post-

CSKP kinematics, core muscle fatigue) to isolate just the effect of utilizing a kinematic compensation strategy on biomechanical variables previously associated with running injuries. Participants with experimentally reduced core stability adopted a kinematic

63 compensation strategy associated with an increased peak patellofemoral joint reaction force (6.2±1.0 BW vs 6.6±0.7 BW, p=0.029), peak knee abduction moment (4.3±0.9

%BW*h vs 5.1±0.8 %BW*h, p=0.01), knee abduction impulse (0.6±0.2 %BW*h*s vs

0.7±0.3 %BW*h*s, p=0.02), and peak knee extension moment (16.4±1.5 %BW*h vs

17.2±2.4 %BW*h, p=0.09) during the stance phase of running. Increases in these variables have previously been associated with patellofemoral pain syndrome and iliotibial band syndrome, two of the most common running injuries. None of these effects were significantly altered by the presence of core muscle fatigue. This study suggests that novice runners with insufficient core stability may adopt altered movement patterns associated with increased lower extremity loading and ultimately a higher risk of sustaining a running-related injury.

4.2 Introduction

Core stability has become an extremely popular topic among physical therapists, physicians, researchers, and physically active people in general (Kibler, Press et al. 2006;

McGill 2010). For the purpose of this study, core stability is defined as the body’s ability to maintain or resume an equilibrium trunk and pelvis position (or trajectory) following internal or external perturbations (Zazulak, Hewett et al. 2007). Core stability is a dynamic property, as the core musculature (muscles of the abdomen and lower back) must constantly react to the varying positions and loading conditions the body experiences during running (Willson, Dougherty et al. 2005). Additionally, many components contribute to one’s ability to stabilize their core. These include core muscle strength, endurance, and neuromuscular coordination. Fatigue of the core musculature

64 may negatively affect one or more of these components of core stability and lead to a reduced ability to control the trunk (Potvin and O'Brien 1998; Paasuke, Ereline et al.

1999; Singh, Arampatzis et al. 2010; Downey, Kamalapurkar et al. 2016).

A runner’s ability to control their trunk will likely have a large influence on the function of the lower extremities and the loads they experience, as over half of the body’s mass resides in the upper body (Dempster and Gaughran 1967). A runner with insufficient control of this mass may have increased loading of the lower extremities and spine during running if this lack of control results in atypical muscle activity or movement of the trunk and/or lower extremities. Additionally, it has been shown that muscle activity of the trunk precedes dynamic movement of the extremities (Hodges and

Richardson 1997; Hodges and Richardson 1997; Jamison, McNally et al. 2013). This research has led people to believe that sufficient core stability allows for the most effective and efficient production, absorption, and control of force and motion to the lower extremities. Naturally arising from this belief is the popular theory that insufficient core stability may lead to less efficient movements and ultimately musculoskeletal injury

(Fredericson and Moore 2005; Willson, Dougherty et al. 2005; Kibler, Press et al. 2006); however, this theory is lacking sufficient scientific evidence.

To our knowledge, there have been no scientific studies that identify how individuals lacking sufficient core stability may compensate for this deficiency during running. It is important to understand strategies that may be utilized by runners to compensate for poor core stability in order to understand the biomechanical consequences associated with this deficiency and what injuries may consequently arise in

65 these runners. This knowledge may ultimately contribute to the development of improved running injury prevention and rehabilitation protocols.

Research has shown people may adopt potentially detrimental movement patterns when stability of a joint or structure is compromised. Patients with poor knee stability following an anterior cruciate ligament (ACL) injury adopted significantly different movement patterns associated with increased stress on the articular cartilage of the knee when compared to those who had excellent knee stability following the injury (Rudolph,

Axe et al. 2001). Patients with ACL deficient knees were found to have different gait kinematics in the deficient limb when compared to their contralateral limb (Andriacchi and Dyrby 2005), and these gait differences predict patterns of cartilage degeneration

(Andriacchi, Briant et al. 2006). Additionally, patients with chronic ankle instability were found to adopt different kinematics during walking that were associated with increased stress on ankle joint structures when compared to healthy matched controls (Monaghan,

Delahunt et al. 2006). Similarly, runners lacking sufficient core stability may alter their movement patterns during running in such a way that could be detrimental if the kinematics adopted have been previously associated with increased injury risk. The adoption of altered movement patterns in the presence of instability may be driven by a number of factors. Energy cost may be one of these driving factors, as the minimization of energy expenditure is believed to play a large part in preferred movement patterns

(Cavanagh and Williams 1982; Alexander 1984; Sparrow 2000).

The purpose of this study was to determine the biomechanical consequences of altering running kinematics in response to reduced core stability using kinematically-

66 driven simulations (Fig. 4.1). Analyses were repeated in two separate cases with the level of core muscle fatigue in the model held constant, either with or without muscle fatigue, in order to isolate only the effect of altering kinematics on biomechanical variables and determine if that effect is dependent on the level of core muscle fatigue (Fig. 4.1A). We hypothesized that altered kinematics would result in increased lower extremity loading, spinal loading and energy consumption during stance without core muscle fatigue and increased lower extremity loading and spinal loading but decreased energy consumption during stance with muscle fatigue (Fig. 4.1B).

Figure 4.1. Diagram of the study’s purpose (A), hypotheses (B), and motivation (C).

4.3 Methods

4.3.1 Participants

Twenty-five novice runners (23.6±6.8 years; 13 female; 1.74±0.08 m; 69.4±12.6 kg) participated in a separate larger study (Chapter 2) after providing IRB-approved informed consent. Participants were recruited from surrounding communities and novice runners were defined as running <10 miles per week on average and not playing running sports (e.g., soccer) more than once a week. Other exclusion criteria for this study were as follows: BMI>30; history of lower back pain; musculoskeletal injury within the past 3 months. Eight of these runners (3F/5M, 26.2±11.1 years; 1.77±0.09 m; 71.6±16.7 kg) were included in this simulation study (Table D.1).

4.3.2 Core Stability Knockdown and Running Biomechanics

Experimental three-dimensional overground jogging kinematics and kinetics were collected before and after participants performed a fatiguing core stability knockdown protocol (CSKP). We previously developed this protocol to reduce a person’s core stability temporarily in a single testing session by fatiguing both the superficial and deep core musculature with minimal involvement from the lower extremity muscles (Table

Participants concurrently performed a secondary task of counting backwards by 4’s to provide a more functional measure of core stability, as real-life situations almost always require attention to be divided between multiple tasks (Moghadam, Ashayeri et al. 2011).

The starting number was incremented by one for each trial. This task isolates control of the lumbar spine and trunk from adjustments in the lower extremity joints. A >10% increase in CoPexc was defined a priori as a meaningful experimental decrease in core stability.

Electromyography (EMG) was used to monitor muscle activity during the CSKP.

Electrodes were placed bilaterally on the external obliques (EO) and internal obliques

(IO), and unilaterally on the dominant-side rectus abdominis (RA), and L5 lumbar extensor (L5) during the CSKP. Dual electrodes were used for the EO, RA and L5 muscles and two single electrode discs were used for the IO muscles (Jamison, McNally et al. 2013). Pre-gelled (Ag/AgCl), surface electrodes (A10011/A10005; Vermed, Inc;

Bellows Falls, VT, USA) were placed as recommended by McGill to best reflect deep muscle activity (McGill, Juker et al. 1996) or directly on the most prominent aspect of the muscle belly and oriented parallel to the muscle fibers. Electrode locations were shaved, if necessary, and cleaned with alcohol pads. The median frequency (MedF) of the raw

EMG signal for all core muscles was measured at the beginning and the end of the isometric exercises during the CSKP in order to obtain a measurement of core muscle fatigue induced by the CSKP. A decrease in the MedF of an EMG signal has been directly related to the level of muscle fatigue (De Luca 1983). A ≥10% decrease in the

MedF has been related to a meaningful level of muscle fatigue (Hart, Fritz et al. 2006;

Hart, Kerrigan et al. 2009).

Marker data were collected at 300Hz using 9 Vicon MX-F40 cameras (Vicon

Motion Systems; Oxford, UK) and filtered using a 4th order Butterworth filter at 15Hz.

Ground reaction forces (GRF) were recorded at 1500Hz from Bertec 4060-10 force plates

(Bertec Corp; Columbus, OH, USA). The speed of each trial was monitored using timing gates (Fusion Sport; Sumner Park, QLD, Australia) to ensure all jogging trials were within ±10% of a self-selected, comfortable jogging speed. A customized set of retro- reflective markers were placed on the upper and lower body (Andriacchi, Alexander et al.

1998; Jamison, Pan et al. 2012) (Fig. 2.1) and ten overground jogging trials were collected for each leg, both pre- and post-CSKP.

The subset of participants chosen for this simulation study had to fit the following requirements in order to test the study’s primary hypotheses: 1) >10% decrease in core stability and/or core muscle fatigue >10% experimentally following the CSKP and 2) significant change in at least one peak hip or knee kinematic variable previously associated with running injuries post-CSKP. A significant change in kinematics was determined from a 2-sided two sample t-test on an individual participant for each kinematic variable, comparing their pre-CSKP data to post-CSKP data. Of the 8 participants included in this study that fit the above criteria, four (0F/4M, 75.0±11.8 kg,

1.8±0.1 m, 21.5±3.3 y) had core muscle fatigue and reduced core stability experimentally following the CSKP (CS group) and four (3F/1M, 68.1±22.0 kg, 1.7±0.1 m, 31.0±14.6 y) had core muscle fatigue only and no change in core stability following the CSKP (NCS group). All analyses were done separately on the two groups to separate the effects that may be driven by a reduction in core stability from ones that may be driven by core muscle fatigue.

For each of the 8 included participants, the experimental marker and force data from the motion capture system from one good pre-CSKP trial and one good post-CSKP trial, where all reflective markers were present and each force plate was contacted by at most one foot at a time, were arbitrarily chosen and used as input into the musculoskeletal modeling software OpenSim to create subject-specific kinematically driven simulations of each participant running. All simulations were performed over the stance phase of the dominant leg (ipsilateral) for each participant.

4.3.3 Simulation procedures

Simulations are a valuable complement to experiments, as they enable the estimation of important variables that cannot be measured experimentally, including forces generated by individual muscles and internal joint loads. Additionally, simulations allow cause-effect relationships in complex dynamic systems to be established since one parameter can be changed while the rest are held constant. In this study, simulations facilitate the estimation of internal lower extremity and spinal loading, energy consumption, and individual muscle force production. They also allow solely the effect of changing running kinematics to be investigated while all other parameters in the model are held constant.

71 latissimus dorsi (LD), quadratus lumborum (QL), multifidus (MF), psoas (PS) and the erector spinae (ES) which is comprised of the superficial and deep longissimus thoracis

(LT) and iliocostalis lumborum (IL)). The lumbar spine muscle groups are modeled as multiple fascicles to account for their large surface area and multiple lines of action. The spine consists of the sacrum, five individual lumbar vertebrae, and lumped thoracic and cervical vertebrae. The FBLS model is freely available on the SimTK website

(https://simtk.org/home/fullbodylumbar). This model does not include the TrA due to modeling limitations.

The workflow used to create each simulation is shown in Figure 1.4. First, the generic musculoskeletal model was first scaled to anthropometrically match the individual participant for whom data had been collected (Delp, Anderson et al. 2007).

Next, motion and mass properties of the model were optimized using inverse kinematics and residual reduction algorithms to achieve a dynamically consistent set of kinematics and kinetics that best matched the experimentally collected data (Delp, Anderson et al.

2007). Next, static optimization (SO) was performed to resolve the net joint moments into individual muscle forces at each instant in time (Steele, Demers et al. 2012). Non- physical forces and moments added to the model to resolve dynamic inconsistencies

(residuals) or assist muscles that are too weak to produce the necessary joint moment

(reserves) were compared to the established thresholds (Figure E.33) to ensure all simulations were successful (Hicks 2011; Lund and Dembia 2013) (Appendix E). Lastly, the JointReaction analysis tool was used to calculate the internal joint loads on the lumbar spine and lower extremities (Steele, Demers et al. 2012).

Figure 4.2. Schematic describing the four subject-specific simulations developed for each participant. Simulation conditions involved two kinematic states (before and after the core stability knockdown protocol (CSKP)) and two core muscle fatigue states (no fatigue and the subject-specific level of fatigue induced by the CSKP).

Four subject-specific simulations were developed for each participant (Fig. 4.2).

A baseline simulation of the stance phase was created for each kinematic state (pre-

CSKP, post-CSKP) and then a fatigued simulation was created for each kinematic state.

Since fatigue is generally defined as the reduced ability of a muscle to produce force, the

m maximum isometric force property (F 0) of all trunk muscles in the model was reduced by a subject-specific level determined from the EMG recorded experimentally during the

CSKP to recreate the muscle fatigue induced in each participant during the CSKP. For

m each participant, the percent change that F 0 was reduced in all the core muscles was

푚 ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ calculated using the following relationship: %∆F0 = (%∆MedF푐표푟푒) ∗ 2.57, where

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ %∆MedF푐표푟푒 is the average percent change in core muscle MedF, which represents the subject-specific level of muscle fatigue induced by the CSKP. The scaling factor of 2.57 was determined through a literature search and takes into account the relationship between the change in a muscle’s force production and the corresponding change in its

MedF (Fig. 4.3) (Roy and De Luca 1989; Allison and Fujiwara 2002; Bilodeau,

Schindler-Ivens et al. 2003; Drechsler, Cramp et al. 2006).

-100 -80 -60 -40 -20 0 0 Roy 1989 -5 -10 Drechsler 2006 -15 Bilodeau 2003 -20 A & F 2002 -25

100%MVIC -30 -35

Percent Change Change in MedF from Percent -40

-45 Percent Change in Muscle Force Production from 100%MVIC

Figure 4.3. Previously recorded data investigating how a muscle’s change in force production is related to a change in its MedF. The a line was fit to the previous data with a forced intercept of 0 to determine the scaling factor for simulating muscle fatigue based off of a change in MedF.

Each study in Figure 4.3 involved participants performing a maximal voluntary isometric contraction (MVIC) for a given muscle and the resulting force production and 74

MedF was measured. The MedF for each muscle was then recorded at different levels of reduced force production in relation to the force produced during the MVIC. A line was fit to all the data in Figure 4.3 with a forced intercept of 0 to determine the scaling factor for simulating muscle fatigue based off of a change in MedF.

4.3.4 Statistics

Statistical analyses were performed in JMP® Pro, Version 12.2.0 (SAS Institute

Inc., Cary, NC). Linear mixed models were used to evaluate the effect of a change in kinematics on the primary variables of interest in the presence of core muscle fatigue or no fatigue. The kinematic state (pre-CSKP, post-CSKP), the core muscle fatigue state

(none, fatigue), and the interaction between kinematic state and fatigue state were treated as fixed effects and participant was treated as a random effect. The best model (based on

AICc and BIC criterion (Burnham and Anderson 2004)) for each of the running parameters was used for analysis. The significance level for all tests was set at α=0.05 and corrections for multiple comparisons were not made due to the exploratory nature of this study. Results found to be trending toward significance (0.05≤p≤0.1) were also reported. Analyses were conducted separately for the CS and NCS groups.

The primary variables of interest investigated were those previously identified to be associated with running injury risk, as well as variables that may be directly related to the compensation strategy utilized. These variables include: peak axial compressive load acting on the tibia (pTibcomp) (Davis, Bowser et al. 2015); peak patellofemoral joint reaction force acting on the patella (pFPFJ) (James 1995; Chen and Powers 2014); internal peak knee abduction (pKAbd) moment and impulse (KAbdI) (James 1995; Stefanyshyn,

Stergiou et al. 2006); internal peak knee extension (pKExt) moment (James 1995;

Salsich, Brechter et al. 2001); internal peak hip abduction (pHAbd) moment and impulse

(HAbdI) (MacMahon, Chaudhari et al. 2000; Willson, Binder-Macleod et al. 2008); total energy consumption (Etot) estimated as total cubed muscle stress

3 푛 3 푛 퐹푚 (퐸푡표푡 ≈ ∑푚=1 (휎푚) ≈ ∑푚=1 ( ) ) (Crowninshield and Brand 1981); peak and 푃퐶푆퐴푚 impulse of the anterior shear and axial compressive loads (Fant_shear, Fcompression,) on each lumbar vertebra (L1-S1); and peak muscle force for the following ipsilateral and contralateral lower extremity and trunk muscles: gluteus maximus (GMax), gluteus medius (GMed), gluteus minimus (GMin), iliacus, rectus femoris (RF), RA, EO, IO, ES

(IL and LT), QL, MF, and PS. All loads reported on a given vertebra are those acting upon it from the superior vertebra (Fig. 4.4). All variables were analyzed over the dominant-side (ipsilateral) stance phase of running only. Lower extremity variables were analyzed only on the ipsilateral limb and trunk muscle forces were analyzed on both the ipsilateral and contralateral side.

Figure 4.4. Diagram of the spinal column showing the vertebral loads analyzed in this study. All loads reported on a given vertebra are those acting upon it from the superior vertebra. For example, as shown in the figure, the L2 loads reported are those from L1 acting on L2. Image adapted from (Neumann 2013).

4.4 Results

4.4.1 Kinematic Changes

Before examining the effect of altering kinematics on biomechanical running variables, it is important to understand what kinematic changes were made in response to reduced core stability following the CSKP. Table 4.1 shows the primary changes in peak joint angles that occurred from pre- to post-CSKP in the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not

(NCS group).

While this study is underpowered to detect significant differences in the variables shown in Table 4.1, the group means in Table 4.1 suggest that the general kinematic strategies adopted by each group may characteristically differ and often are opposite. For example the CS group had an increased peak trunk bending angle (31.9% increase), peak hip extension angle (11.4% increase), and peak hip adduction angle (27% increase) while decreases in all these variables were seen in the NCS group (20.2%, 9.7%, 11.1% decreases, respectively). Therefore these data suggest that while both groups are adopting different kinematics following the CSKP, the group that had reduced stability post-CSKP may be utilizing a different kinematic strategy than the group that was able to maintain their stability following the fatiguing protocol.

CS group NCS group Joint Angle (°) (mean ± SD) (mean ± SD) Trunk Extension (+) pre-CSKP -3.37±6.7 0.5±8.3 post-CSKP -2.4±6.7 -1.7±4.6 Trunk Flexion (-) pre-CSKP -20.5±9.8 -10.57±6.7 post-CSKP -20.3±8.1 -13.25±2.8

Trunk Ipsilateral Bending (+) pre-CSKP 9.4±2.7 8.4±4.3 post-CSKP 12.4±2.1 6.7±5.8 Trunk Contralateral Rotation (+) pre-CSKP 41.7±15.9 25.1±10.3 post-CSKP 39.67±12.1 32.0±17.0

Hip Extension (-) pre-CSKP -14.9±5.6 -15.95±5.5 post-CSKP -16.6±6.3 -14.4±2.4 Hip Adduction (+) pre-CSKP 11.1±6.5 14.4±5.5 post-CSKP 14.1±4.4 12.8±3.5 Table 4.1. Peak joint angles during running before and after the core stability knockdown protocol (CSKP) for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group). Joint angles shown are those that were the most different between conditions. The (+) or (-) with each variable indicates which direction corresponds to an increase in the given joint angle based on joint angle definitions.

4.4.2 Effect of a Kinematic Compensation Strategy

Tables 4.2, 4.3, and 4.4 show the kinematic effect results from linear mixed- model analyses for the biomechanical variables of interest. Analyses were performed on both the group that experienced a meaningful reduction in core stability experimentally post-CSKP (CS group, N=4) and the group that did not (NCS group, N=4) in order to

79 provide insight into which effects may be ultimately driven by a reduction in core stability. Results are presented for only the variables that were significantly or trending towards being significantly affected by kinematics. Differences between the variables of interest in the pre-CSKP kinematic state and the post-CSKP kinematic state and their mixed-model kinematic effect p-values are shown in Tables 4.2-4.4 and the full model statistics along with group means for each condition (pre-CSKP, no fatigue; pre-CSKP, fatigue; post-CSKP, no fatigue; post-CSKP, fatigue) can be found in Tables C.1-C.3 in

Appendix C. Significant p-values in Tables 4.2-4.4 indicate that a change in kinematics following the CSKP had a significant effect on the corresponding biomechanical variables. No kinematic effects were found to be significantly dependent on the presence of simulated core muscle fatigue (p>0.05 for all kinematic*fatigue interaction effects).

In the CS group, a change in kinematics in response to reduced core stability was associated with reduced energy consumption (p=0.059) and increased peak knee abduction moment (pKAbdM) (p=0.01), knee abduction impulse (KAbdI) (p=0.02), peak patellofemoral joint reaction force (pFPFJ) (p=0.029), and peak knee extension moment

(pKExtM) (p=0.09) during stance (Table 4.2). None of these same effects were seen in the NCS group, suggesting these biomechanical changes may be ultimately driven by this group having experimentally reduced core stability following the CSKP.

Table 4.2. Differences in energy consumption and lower extremity loading variables from the pre-CSKP kinematic state to the post-CSKP kinematic state and the corresponding mixed model kinematic effect p-values. Analyses were performed separately for the group that experienced a meaningful reduction in core stability experimentally post- CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be driven by a reduction in core stability. Variables defined as larger in the negative direction are indicated with (-). CS Group NCS p-value p-value (post-pre) Group (post-pre) Energy [MPa3] -239.38 0.059 74.06 0.508

pFPFJ [BW] 0.38 0.029 -0.06 0.856

pTib (-) comp -0.30 0.17 -0.44 0.08 [BW]

pKabdM (-) -0.74 0.01 0.68 0.003 [%BW*h]

KAbdI (-) -0.10 0.02 0.10 0.0005 [%BW*h*s]

pKExtM 0.76 0.09 -0.48 0.58 [%BW*h]

Abbreviations: peak patellofemoral joint reaction force (pFPFJ), peak tibial compression (pTib comp), peak knee abduction moment (pKAbdM), knee abduction impulse (KAbdI), peak knee extension moment (pKExtM). All forces, moments, and impulses are internal. Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are italicized. Forces are normalized to body weight (BW) and moments and impulses are normalized to BW*height (BW*h).

Figure 4.5 shows the group means for each simulation condition in both the CS and NCS groups for the pFPFJ and KAbdI results. The p-value with each group corresponds to the test of the kinematic effect from the linear mixed models. The

81 increases in lower extremity loading experienced in the CS group following a change in kinematics are evident in Figure 4.5 and it is clear that there is no effect of simulated core muscle fatigue on the pFPFJ and KAbdI, as the ‘fatigue’ and ‘no fatigue’ conditions are entirely overlapping. This same result was seen with all lower extremity loading variables. Individual changes in the pFPFJ from pre- to post-CSKP in the CS group over the entire stance phase of running are shown in Figure 4.6. Fatigue had no effect on pFPFJ magnitudes, therefore the loading shown in Figure 4.6 is representative of both the

‘fatigue’ and ‘no fatigue’ simulation conditions.

Figure 4.5. The effect of altered kinematics (‘pre’ or ‘post’ CSKP) on the peak patellofemoral joint reaction force (JRF) and knee abduction impulse during the stance phase of running.Group means and standard deviations for each simulation condition are shown in both the group that had reduced core stability in the ‘post’ kinematic condition (CS group) and the group that did not (NCS group). The kinematic effect for each group is investigated in two fatigue conditions of the core musculature (no fatigue, fatigue). The p-value with each group corresponds to the test of the kinematic effect from linear mixed models. Knee abduction is defined as the negative frontal plane angle.

Figure 4.6. Individual changes in the ipsilateral patellofemoral joint reaction force (JRF) during stance from pre- to post-CSKP in the CS group in units of body weight (BW). Fatigue had no effect on patellofemoral JRF magnitudes, therefore the loading shown is representative of both the ‘fatigue’ and ‘no fatigue’ simulation conditions.

Cumulative lumbar spinal compression during the stance phase of running in the

CS group was significantly reduced by a change in running kinematics (Table 4.3). A change in kinematics in this group was associated with reduced spinal compressive force impulses on the following lumbar vertebrae: L1 (p=0.06), L2 (p=0.049), L3 (p=0.046),

L4 (p=0.04), L5(p=0.04), S1 (p=0.04). These results were not seen in the NCS group,

84 suggesting that the kinematic compensation strategy adopted in response to reduced core stability is associated with reduced cumulative compression on the lumbar spine.

Table 4.3. Differences in spinal loading variables from the pre-CSKP kinematic state to the post-CSKP kinematic state and the corresponding mixed model kinematic effect p- values. Analyses were performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be ultimately driven by a reduction in core stability. All loads reported on a given vertebra are those acting upon it from the superior vertebra. NCS CS Group p-value Group p-value (post-pre) (post-pre)

impL1 Comp -3.2 0.06 0.62 0.75 [%BW*s] 0

impL2 Comp -3.46 0.049 0.70 0.74 [%BW*s]

impL3 Comp -4.0 0.046 0.94 0.67 [%BW*s]

impL4 Comp -4.34 0.04 1.0 0.67 [%BW*s]

impL5 Comp -4.40 0.04 0.86 0.72 [%BW*s]

impS1 Comp -4.20 0.04 0.78 0.73 [%BW*s]

Abbreviations: impulse of compressive load acting on S1 (impS1comp). All loads are internal. Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are italicized. All spinal loading variables are normalized to body weight (BW).

In general, a change in kinematics following the CSKP in the CS group was associated with an increase in the peak force production of select lower extremity muscles and a decrease in the peak force production of select core muscles (Table 4.4).

The kinematic state (pre- or post-CSKP) of runners in the CS group had a significant or trending positive effect on the gluteus minimus (GMin) (p=0.001) and rectus femoris

(RF) (p=0.004) peak muscle force and a negative effect on the ipsilateral gluteus maximus (iGMax) (p=0.07), contralateral psoas (cPS) (p=0.06) and the contralateral internal oblique (cIO) (p=0.027). Since these effects were not also observed in the NCS group, this further supports that these results may be directly related to a reduction in core stability following the CSKP.

Table 4.4. Differences in peak muscle force variables from the pre-CSKP kinematic state to the post-CSKP kinematic state and the corresponding mixed model kinematic effect p- values. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be ultimately driven by a reduction in core stability.

CS Group NCS Group p-value p-value (post-pre) (post-pre) iGMax m -3.46 0.07 1.38 0.7 [%F 0] iGMed m 5.48 0.12 -4.42 0.0002 [%F 0] iGMin m 22.18 0.001 -10.78 0.026 [%F 0] iIliacus m 8.32 0.15 -1.66 0.02 [%F 0] iRF m 4.46 0.004 -0.06 0.99 [%F 0] cPS m -0.64 0.06 -0.42 0.35 [%F 0] cIO m -4.46 0.027 -0.60 0.84 [%F 0] Abbreviations: ispilateral side (i), contralateral side (c), gluteus maximus (GMax), gluteus medius (GMed), gluteus minimus (GMin), rectus femoris (RF), psoas (PS), and internal oblique (IO). Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are bolded and italicized. All m muscle forces are normalized to the muscle group’s maximum isometric force property (F 0).

4.5 Discussion

Using subject-specific simulations, this study found that a group of runners that experimentally experienced a meaningful reduction in core stability following a core muscle fatiguing protocol (CS group) adopted a different kinematic compensation strategy than a group of runners that was able to maintain their core stability in the presence of muscle fatigue (NCS group). The kinematic strategy adopted by the CS group was trending or significantly associated with decreased energy consumption, cumulative lumbar spinal compression, and core muscle force production and increased

87 lower extremity loading and muscle force production during the stance phase of running.

These associations were not observed in the NCS group, further supporting the theory that these associations may be the result of a kinematic strategy specifically adopted as a result of this group having reduced core stability. These relationships were not affected by the level of core muscle fatigue simulated in the model, suggesting the consequences of changing kinematics are similar regardless of core muscle fatigue.

These results suggest that a group of runners with reduced core stability may choose to alter their kinematics to compensate because this strategy reduces their energy consumption, core muscle force production, and compressive loading on the spine.

However, detrimental consequences are also associated with this compensation strategy.

These consequences include: 1) increased lower extremity loading parameters that have previously been associated with increased risk of developing two of the most common running injuries, patellofemoral pain syndrome (PFPS) and iliotibial band syndrome

(ITBS) and 2) an increased workload of the RF and GMin muscles, increasing the risk of injury or fatigue in the muscles during running.

An increased internal knee extension moment (pKExtM) during running is suggestive of an increased load on the patella, as this moment is directly related to the quadriceps workload. An increase in the quadriceps workload directly leads to an increase in the tensile force of the quadriceps tendon and patellar ligament acting on the patella (Andriacchi, Kramer et al. 1985; James 1995). In fact, these relationships were confirmed as an increase in the RF peak muscle force and an increase in the peak patellofemoral joint reaction force were observed in the CS group along with the

88 increased pKExtM. Increased loading on the patella is believed to increase risk for developing PFPS (James 1995). While the threshold for increased patellar loading that elicits PFPS remains unknown, there is in-vitro evidence that increased repetitive stress placed on the patella accelerates retropatellar cartilage damage (Zimmerman, Smith et al.

1988). Heino Brechter and Powers used a biomechanical model to estimate a 2x greater peak patellofemoral joint stress during walking in participants with PFP versus controls

(6.61 MPa vs 3.13 MPa) (Heino Brechter and Powers 2002).

An increased internal peak knee abduction moment (pKAbdM) and internal knee abduction impulse (KAbdI) have previously been associated with both increased PFPS

(Stefanyshyn, Stergiou et al. 2006) and ITBS injury risk (James 1995). Stefanyshyn et al. followed a group of 80 healthy runners over a 6-month training period and found that the runners who developed PFP over this time period had a significantly larger KAbdI during stance than matched runners who remained healthy (9.2±3.7 Nms vs 4.7±3.5 Nms)

(Stefanyshyn, Stergiou et al. 2006). The internal pKAbdM during running is directly related to the load placed on the lateral stabilizing structures of the knee. Therefore, an increase in this moment likely is representative of increased tensile strain placed on the iliotibial band (ITB) (Powers 2010), which may increase risk of developing ITBS

(Andriacchi, Kramer et al. 1985; James 1995; Fairclough, Hayashi et al. 2006).

This study suggests that novice runners with insufficient core stability may adopt a kinematic compensation strategy associated with increased lower extremity loading and ultimately a higher risk of sustaining a running-related injury. This work provides support for future research to investigate interventions designed to improve core stability to

89 determine if increasing core stability may result in runners adopting “lower-risk” running mechanics.

4.5.1 Limitations

Several limitations should be kept in mind when interpreting these results. First, the FBLS model does not have degrees of freedom in the knee in the frontal or transverse planes. Knee internal rotation and knee abduction are kinematic parameters that have both been previously associated with running injuries like PFPS and ITBS (Huberti and

Hayes 1984; Powers 2003; Noehren, Davis et al. 2007; Ferber, Noehren et al. 2010).

Although the motion of the knee in these planes is generally quite small, especially during running, and the simplification of the knee as a hinge joint likely will not greatly affect the results of this study, modeling these parameters in the future may provide further insight into the effects of a kinematic compensation strategy on lower extremity loading. Despite the fact that these degrees of freedom in the knee are not modeled, internal joint loads for all degrees of freedom are still estimated in OpenSim’s

JointReaction Analysis tool. The internal loads estimated by the JointReaction Analysis correspond to the internal loads carried by all un-modeled joint structures (i.e., cartilage, ligaments, etc.) in order to produce the desired kinematics (Steele, Demers et al. 2012).

Therefore, these internal moments, like the internal knee abduction moment (kAbdM) for example, can be interpreted as the load that would be carried by the lateral structures of the knee preventing the knee from moving into any knee adduction.

When separating the participant group into the CS and NCS groups, there were no females in the CS group and 3 out of 4 participants in the NCS group were female. While

90 all participants were healthy and joint loads were normalized to body weight and height, this discrepancy should still be acknowledged as it may have affected the study results.

Energy consumption in this study was estimated as the sum of total cubed muscle stress (Crowninshield and Brand 1981). This estimate is entirely dependent on simulated muscle force production. For the purposes of this study, this estimation was sufficient to begin to explore how energy consumption during running may be affected by muscle fatigue and a neuromuscular compensation strategy. Future work specifically interested in energy cost during running may consider using forward dynamic simulations, allowing for more complex models of energy expenditure to be utilized (Umberger, Gerritsen et al.

2003).

Additionally, although there are multiple participants in each analysis group (CS group and NCS group), the sample size in this study is relatively small. While within- subject analyses allow a smaller sample size to be utilized, a larger group of participants may have allowed for further characterization of different kinematic strategies that may be adopted in the presence of muscle fatigue and/or reduced core stability.

4.6 Conclusion

A kinematic compensation strategy adopted in response to an experimentally- induced reduction in core stability was associated with increased knee loading during the stance phase of running that has previously been associated with increased PFPS and

ITBS injury risk. A reduction in energy consumption, spinal loading, and the workload of the core musculature may be ultimately driving the adoption of this compensation strategy.

Chapter 5 : Simulating the Utilization of a Neuromuscular Compensation Strategy in Response to Core Muscle Fatigue during Running

5.1 Abstract

Muscle strength, muscle endurance, and neuromuscular coordination are all components that contribute to a person’s ability to control their trunk during dynamic tasks like running. Fatigue of the core musculature may negatively affect one or more of these components of core stability. Understanding the strategies utilized by runners to compensate for poor core stability will provide insight into the biomechanical consequences of this deficiency and its influence on injury risk. The purpose of this study was to determine the effect of altering only neuromuscular control strategies during running in response to simulated core muscle fatigue using kinematically-driven simulations. Eight novice runners jogged overground in a motion capture laboratory before and after performing a fatiguing core stability knockdown protocol (CSKP).

Subject-specific kinematically-driven jogging simulations were developed for each participant in four different conditions (pre-CSKP kinematics, no core muscle fatigue; pre-CSKP kinematics, core muscle fatigue; post-CSKP kinematics, no core muscle fatigue; post-CSKP kinematics, core muscle fatigue). A neuromuscular compensation

92 strategy adopted in response to simulated core muscle fatigue was associated with decreased lumbar spine compression (p≤0.08) and decreased superficial core muscle force production (p≤0.033). Although between-participant variability prevented significant muscular compensations from being detected, increased deep core muscle force production was observed as the primary compensation utilized in response to simulated overall core muscle fatigue. A neuromuscular compensation strategy was not associated with a significant change in estimated energy consumption or lower extremity loading during the stance phase of running, suggesting this strategy may not affect energy cost or lower extremity loading and injury risk. Only an increase in force production of the deep core musculature was required to execute this compensation strategy, suggesting a neuromuscular training program emphasizing engagement and force production of the deep core muscles may give runners the ability to utilize potentially lower-risk compensation strategies in the presence of core muscle fatigue and/or poor core stability.

5.2 Introduction

The belief that insufficient core stability may lead to less efficient movements and ultimately musculoskeletal injury has become widespread despite minimal scientific evidence supporting this theory. There has been some research linking poor core stability to lower extremity injuries, such as anterior cruciate ligament (ACL) tears and ankle sprains (Bullock-Saxton, Janda et al. 1994; Dempsey, Lloyd et al. 2007; Zazulak, Hewett et al. 2007; Zazulak, Hewett et al. 2007; Donnelly, Lloyd et al. 2012; Jamison, Pan et al.

2012), however the relationship between core stability and running-related injuries remains unclear. A better understanding of the role that core stability plays in running

93 and the influence of core stability on running injury risk is necessary to improve running injury prevention and rehabilitation programs.

For the purpose of this study, core stability is defined as the body’s ability to maintain or resume an equilibrium trunk and pelvis position (or trajectory) following internal or external perturbations (Zazulak, Hewett et al. 2007). Core stability is a dynamic property (Willson, Dougherty et al. 2005) and many components contribute to one’s ability to stabilize their core, including core muscle strength, endurance, and neuromuscular coordination. Fatigue of the core musculature may negatively affect one or more of these components of core stability and lead to a reduced ability to control the trunk (Parnianpour, Nordin et al. 1988; Potvin and O'Brien 1998; Granata and Gottipati

2008; Downey, Kamalapurkar et al. 2016).

Since over half of the body’s mass resides in the upper body (Dempster and

Gaughran 1967), poor control of this mass during running may lead to altered movement patterns and/or increased loading on the lower extremities and the spine (Willson,

Dougherty et al. 2005). Research has shown patients may adopt detrimental movement patterns when stability of a joint like the knee or ankle is compromised (Rudolph, Axe et al. 2001; Monaghan, Delahunt et al. 2006). The adoption of altered movement patterns in the presence of instability may be driven by a number of factors, such as a reduction in energy expenditure (Cavanagh and Williams 1982; Alexander 1984; Sparrow 2000) or a protective attempt to limit loading of the compromised joint (Monaghan, Delahunt et al.

2006).

To our knowledge, there have been no scientific studies that identify how individuals lacking sufficient core stability may compensate for this deficiency during running. A neuromuscular compensation strategy, in which a runner alters only muscular activation strategies rather than movement patterns, may be one type of strategy utilized by runners to compensate for a deficiency in core stability. It is important to identify different strategies that may be utilized by runners to compensate for poor core stability in order to understand the advantages and disadvantages of these strategies, as well as what injuries may arise in runners with this deficiency.

Research has shown that participants could maintain a constant trunk extension torque output in the presence of trunk extensor muscle fatigue by altering the activity of other trunk muscles during the task (Sparto, Parnianpour et al. 1997). Additionally, Gates and Dingwell found that participants were able to alter their control strategy and maintain shoulder movement stability in the presence of shoulder muscle fatigue (Gates and

Dingwell 2010). Therefore, altering neuromuscular control strategies of the trunk in response to core muscle fatigue and/or compromised core stability may be beneficial in order to maintain desired movement patterns. Nevertheless, there may also be detrimental consequences associated with this type of strategy, such as altered spinal loading and increased energy consumption (Sparto, Parnianpour et al. 1997; Granata, Slota et al.

2004; van der Krogt, Delp et al. 2012).

The purpose of this study was to determine the biomechanical consequences of altering neuromuscular control strategies during running in response to core muscle fatigue using kinematically-driven simulations (Fig. 5.1). Analyses were repeated in two

95 separate kinematic states, corresponding to participants’ kinematic state either before or after performing an experimental fatiguing core stability knockdown protocol (CSKP)

(Fig. 5.1A). By simulating fatigue and holding kinematics constant, we isolated the effect of altering only neuromuscular control strategies on biomechanical variables and determined if that effect is dependent on a runner’s kinematic state. We hypothesized that utilizing a neuromuscular compensation strategy for simulated core muscle fatigue would result in increased energy consumption, lower extremity loading, and spinal loading in both kinematic states, with larger increases seen in the pre-CSKP state (Fig.

5.1B).

Figure 5.1. Diagram of the study’s purpose (A), hypotheses (B), and motivation (C).

5.3 Methods

5.3.1 Participants

Twenty-five novice runners (23.6±6.8 years; 13 female; 1.74±0.08 m; 69.4±12.6 kg) participated in a separate larger study (Chapter 2) after providing IRB-approved informed consent. Participants were recruited from surrounding communities and novice runners were defined as running <10 miles per week on average and not playing running sports (e.g., soccer) more than once a week. Other exclusion criteria for this study were as follows: BMI>30; history of lower back pain; musculoskeletal injury within the past 3 months. Eight of the 25 runners (3F/5M, 26.2±11.1 years; 1.77±0.09 m; 71.6±16.7 kg) were included in this simulation study (Table D.1).

5.3.2 Core Stability Knockdown and Running Biomechanics

Three-dimensional overground jogging kinematics and kinetics were collected experimentally before and after participants performed a novel core stability knockdown protocol (CSKP). We previously developed this protocol to reduce a person’s core stability temporarily in a single testing session by fatiguing both the superficial and deep core musculature with minimal involvement from the lower extremity muscles (Table

2.1). Core stability was assessed before and after this protocol using an unstable quiet sitting test (QST) that measures the center of pressure excursion (CoPexc) while the participant attempts to keep the trunk as still as possible for 60 seconds (Figure 2.2). This task isolates control of the lumbar spine and trunk from adjustments in the lower extremity joints. A >10% increase in CoPexc was defined a priori as a meaningful decrease in core stability.

Electromyography (EMG) was used to monitor muscle activity during the CSKP.

Electrodes were placed bilaterally on the external obliques (EO) and internal obliques

MedF has been related to a meaningful level of muscle fatigue (Hart, Fritz et al. 2006;

Hart, Kerrigan et al. 2009).

Marker data were collected at 300Hz using 9 Vicon MX-F40 cameras (Vicon

Motion Systems; Oxford, UK) and filtered using a 4th order zero-phase-lag low-pass

Butterworth filter at 15Hz. Ground reaction forces (GRF) were recorded at 1500Hz from

Bertec 4060-10 force plates (Bertec Corp; Columbus, OH, USA). The speed of each trial was monitored using timing gates (Fusion Sport; Sumner Park, QLD, Australia) to ensure all jogging trials were within ±10% of a self-selected, comfortable jogging speed. A

98 customized set of retro-reflective markers were placed on the upper and lower body

(Andriacchi, Alexander et al. 1998; Jamison, Pan et al. 2012; Thompson, Chaudhari et al.

2013) (Fig. 2.1) and ten overground jogging trials were collected for each leg, both pre- and post-CSKP.

2-sided two sample t-test on an individual participant for each kinematic variable, comparing their pre-CSKP data to post-CSKP data. Of the 8 participants included in this study that fit the above criteria, four (0F/4M, 75.0±11.8 kg, 1.8±0.1 m, 21.5±3.3 y) had core muscle fatigue and reduced core stability experimentally following the CSKP (CS group) and four (3F/1M, 68.1±22.0 kg, 1.7±0.1 m, 31.0±14.6 y) had core muscle fatigue only and no change in core stability following the CSKP (NCS group). All analyses were done separately on the two groups to separate the effects that may be driven by core stability from ones driven by core muscle fatigue.

99 driven simulations of each participant running. All simulations were performed over the stance phase of the dominant leg (ipsilateral) for each participant.

5.3.3 Simulation procedures

Simulations are a valuable complement to experiments, as they enable the estimation of important variables that cannot be measured experimentally, including forces generated by individual muscles and internal joint loads. Additionally, simulations allow cause-effect relationships in complex dynamic systems to be established since one parameter can be changed while the rest are held constant. In this study, simulations facilitated the estimation of internal lower extremity and spinal loading, energy consumption, and individual muscle force production. They also allowed the investigation of the isolated effect of simulated core muscle fatigue and a subsequently adopted neuromuscular compensation strategy while running kinematics were held constant.

The full-body lumbar spine (FBLS) model (Fig. 3.1) with 21 segments, 29 degrees-of-freedom, and 324 musculotendon actuators was used to create each running simulation (Raabe and Chaudhari 2016). This model is the first OpenSim full-body model to include the major muscle groups of the lower extremities in addition to the eight major muscle groups of the lumbar spine (RA, EO, IO, latissimus dorsi (LD), quadratus lumborum (QL), multifidus (MF), psoas (PS) and the erector spinae (ES) which is comprised of the superficial and deep longissimus thoracis (LT) and iliocostalis lumborum (IL)). The lumbar spine muscle groups are modeled as multiple fascicles to account for their large surface area and multiple lines of action. The spine consists of the

100 sacrum, five individual lumbar vertebrae, and lumped thoracic and cervical vertebrae.

The FBLS model is freely available on the SimTK website

(https://simtk.org/home/fullbodylumbar). This model does not include the TrA due to modeling limitations.

The workflow used to create each simulation is shown in Figure 1.4. First, the generic musculoskeletal model was scaled to anthropometrically match the individual participant for whom data had been collected (Delp, Anderson et al. 2007). Next, motion and mass properties of the model were optimized using inverse kinematics and residual reduction algorithms to achieve a dynamically consistent set of kinematics and kinetics that best matched the experimentally collected data (Delp, Anderson et al. 2007). Next, static optimization (SO) was performed to resolve the net joint moments into individual muscle forces at each instant in time (Steele, Demers et al. 2012). Non-physical forces and moments added to the model to resolve dynamic inconsistencies (residuals) or assist muscles that are too weak to produce the necessary joint moment (reserves) were compared to the established thresholds (Figure E.33) to ensure all simulations were successful (Hicks 2011; Lund and Dembia 2013) (Appendix E). Lastly, the JointReaction analysis tool was used to estimate the internal joint loads on the lumbar spine and lower extremities (Steele, Demers et al. 2012).

Four subject-specific simulations were developed for each participant (Fig. 4.2).

A baseline simulation of the stance phase was created for each kinematic state (pre-

CSKP, post-CSKP) and then a fatigued simulation was created for each kinematic state.

Since fatigue is generally defined as the reduced ability of a muscle to produce force, the

101

m maximum isometric force property (F 0) of all trunk muscles in the model was reduced by a subject-specific level determined from the EMG recorded during the CSKP to simulate the experimental level of muscle fatigue induced in each participant during the

m CSKP. For each participant, the percent change that F 0 was reduced for all core muscles

푚 ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ was calculated using the following relationship: %∆F0 = (%∆MedF푐표푟푒) ∗ 2.57, where

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ %∆MedF푐표푟푒 is the average percent change in core muscle MedF, which represents the subject-specific level of muscle fatigue induced by the CSKP. The scaling factor of 2.57 was empirically determined from published data and takes into account the relationship between the change in a muscle’s force production and the corresponding change in its

MedF (Fig. 4.3) (Roy and De Luca 1989; Allison and Fujiwara 2002; Bilodeau,

Schindler-Ivens et al. 2003; Drechsler, Cramp et al. 2006).

5.3.4 Statistics

Statistical analyses were performed in JMP® Pro, Version 12.2.0 (SAS Institute

Inc., Cary, NC). Linear mixed models were used to evaluate the effect of simulated core muscle fatigue on the primary variables of interest in both the pre-CSKP kinematic state and the post-CSKP kinematic state. The kinematic state (pre-CSKP, post-CSKP), the core muscle fatigue state (none, fatigue), and the interaction between kinematic state and fatigue state were treated as fixed effects and participant was treated as a random effect.

The best model (based on AICc and BIC criterion (Burnham and Anderson 2004)) for each of the running parameters was used for analysis. Corrections for multiple comparisons were not made due to the exploratory nature of this study. Due to the small sample size and exploratory nature of this study, results found to be trending toward

102 significance (0.05≤p≤0.1), as well as some non-significant or trending results, were also reported which may have meaningful clinical implications. These analyses were conducted separately for the CS and NCS groups in order to provide insight into which effects may be ultimately driven by an experimental reduction in core stability.

The primary variables of interest investigated were those previously identified to be associated with running injury risk, as well as variables that may be directly related to a compensation strategy. These variables include: peak axial compressive load acting on the tibia (pTibcomp) (Davis, Bowser et al. 2015); peak patellofemoral joint reaction force acting on the patella (pFPFJ) (James 1995; Chen and Powers 2014); internal peak knee abduction (pKAbd) moment and impulse (KAbdI) (James 1995; Stefanyshyn, Stergiou et al. 2006); internal peak knee extension (pKExt) moment (James 1995; Salsich, Brechter et al. 2001); internal peak hip abduction (pHAbd) moment and impulse (HAbdI)

(MacMahon, Chaudhari et al. 2000; Willson, Binder-Macleod et al. 2008); total energy consumption (Etot) estimated as total cubed muscle stress

(IL and LT), QL, MF, and PS. All loads reported on a given vertebra are those acting upon it from the superior vertebra (Fig. 4.4). All variables were analyzed over the stance

103 phase of running only. Lower extremity variables were analyzed only on the ipsilateral limb and trunk muscle forces were analyzed on both the ipsilateral and contralateral side.

5.4 Results

Tables 5.1 and 5.2 show the fatigue effect results and their corresponding p-values from linear mixed-model analyses for the biomechanical variables of interest. The full model statistics along with group means for each condition (pre-CSKP, no fatigue; pre-

CSKP, fatigue; post-CSKP, no fatigue; post-CSKP, fatigue) can be found in Tables C.1-

C.3 in Appendix C. Significant p-values in Tables 5.1 and 5.2 indicate that simulated core muscle fatigue and the subsequently adopted neuromuscular compensation strategy had a significant effect on the corresponding biomechanical variables. No fatigue effects were found to be significantly affected by a change in kinematics ((p>0.1) for all kinematic*fatigue interaction effects). The level of muscle fatigue experienced following the CSKP was not significantly different between the CS group and NCS group

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ( %∆MedF푐표푟푒 : -18.2±9.1% and -26.5±6.2%, respectively; p=0.19).

In contrast to our hypotheses, simulated core muscle fatigue during running was significantly associated with decreased rather than increased spinal loading. Core muscle fatigue had a significant or trending negative effect on peak lumbar spinal compression during the stance phase of running in only the CS group and on cumulative lumbar spinal compression in both the CS and NCS groups (Table 5.1). These results suggest that decreased peak compressive loading on the lumbar spine may be driven partially by some factor related to the CS group having reduced core stability following the CSKP, as this change was seen in only the CS group and not the NCS group. However it may also be

104 likely that this result is driven by more variability in the NCS group preventing a significant difference in peak spinal compression from being found. With only four participants in each group, even a slight increase in variability in this group could prevent significant differences from being detected.

105

Table 5.1. Differences in spinal loading variables from the baseline (no core muscle fatigue) state to the fatigued state and the corresponding mixed model fatigue effect p-values. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be ultimately driven by a reduction in core stability. All loads reported on a given vertebra are those acting upon it from the superior vertebra.

CS Group NCS Group p-value p-value (fatigue-baseline) (fatigue-baseline)

L1 Comp Peak [%BW] -42.68 0.007 -15.92 0.49 0 Impulse [%BW*s] -3.60 0.036 -3.8 0.078

L2Comp Peak [%BW] -41.06 0.013 -20.92 0.43

106 Impulse [%BW*s] -3.72 0.038 -4.08 0.076 L3 Comp Peak [%BW] -41.06 0.025 -31.88 0.31 Impulse [%BW*s] -3.94 0.06 -4.54 0.08

L4Comp Peak [%BW] -39.96 0.037 -35.78 0.31 Impulse [%BW*s] -3.94 0.06 -4.40 0.076

L5Comp Peak [%BW] -37.90 0.047 -35.78 0.31 Impulse [%BW*s] -3.84 0.06 -4.58 0.08

S1Comp Peak [%BW] -37.60 0.04 -35.40 0.30 Impulse [%BW*s] -3.80 0.06 -4.40 0.08 Abbreviations: impulse of compressive load acting on S1 (impS1comp), peak compressive load acting on S1 (pS1comp). All loads are internal. Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are italicized. All spinal loading variables are normalized to body weight (BW).

Core muscle fatigue was significantly associated with decreased superficial core muscle force production in both the CS and NCS groups. Specifically, the ipsilateral side

(i) and contralateral side (c) RA and IO and the iEO all had significantly decreased peak force production during the stance phase of running in both groups (Table 5.2). In contrast to the superficial core muscles, the deep core muscles increased their peak force production after simulated core muscle fatigue (Fig. 5.2), despite the fact that these muscles were also fatigued, indicating the deep core muscles may be acting as compensators allowing runners to maintain their kinematics in the presence of fatigue of all the core musculature. Different deep core muscles were utilized as compensators in the group that had reduced core stability following the CSKP compared to the group that

m m did not. In the CS group, the cMF (10.3±6.0 %F 0 vs 14.1±9.1 %F 0, p=0.18) and cQL

m m (3.8±2.0 %F 0 vs 7.2±6.8 %F 0, p=0.18) increased their force production and in the NCS

m m m group, the iQL (4.2±3.8 %F 0 vs 8.2±7.9 %F 0, p=0.21) and cPS (0.6±0.5 %F 0 vs

m 1.2±1.6 %F 0, p=0.18) were the primary compensators (Fig 5.2).

Although these changes in deep core muscle force production are not significant or trending towards significance, they are worth noting as they appeared to be the only muscles functioning as compensators in the presence of simulated core muscle fatigue.

When investigating these compensations further, it was evident there was considerable variation among participant responses to simulated core muscle fatigue. An example of these compensations and this variability is presented in Figure 5.3, which shows the individual changes in the cMF force production with core muscle fatigue over the entire stance phase of running in the CS group. Two of the four participants in this group had

107 large increases in peak cMF force production (P88, P46), the third had a slight increase

(P42), and the fourth decreased cMF force production following core muscle fatigue

(P51) (Fig 5.3). Notably, the participant that only had a very slight increase in cMF force production also had the lowest level of core muscle fatigue (Table D.1) and the participant with decreased cMF force production was using this muscle considerably less than the other three participants. Between-participant variability in muscle compensations may be a result of varying levels of core muscle fatigue, as well as variations in running kinematics that may have led to differences in recruitment of the deep core muscles during running.

Figure 5.4 further investigates the individual changes in the peak force production of the deep core muscles that were identified in the CS group as potential compensators for overall core muscle fatigue. Generally, an individual’s compensation strategy was similar for both the cMF (Fig 5.4A) and the cQL (Fig 5.4B) muscles. Additionally, for both compensating muscles, 3 of the 4 participants in the CS group adopted similar compensation strategies in both their pre- and post-CSKP kinematic states. Only one participant (P46) increased force production of the deep core muscles to compensate for overall core muscle fatigue in their pre-CSKP kinematic state, while force production of these muscles decreased in response to fatigue in their post-CSKP kinematic state (Fig

5.4).

Core muscle fatigue did not significantly affect any of the remaining biomechanical variables of interest. Also in contrast to our hypotheses, there was not a

108 significant change in lower extremity loading or estimated energy consumption when utilizing a neuromuscular compensation strategy in the presence of core muscle fatigue.

Figure 5.2. Deep core muscles utilized as compensators in response to core muscle fatigue during running. Compensations are shown for both the group that had experimentally reduced core stability following the CSKP (CS group) compared to the group that did not (NCS group). All participants were right-side dominant.

109

Table 5.2. Differences in energy consumption and peak muscle force variables from the baseline (no core muscle fatigue) state to the fatigued state and the corresponding mixed model fatigue effect p-values. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be ultimately driven by a reduction in core stability.

CS Group NCS Group (fatigued- p-value p-value (fatigued-baseline) baseline) Energy -149.05 0.21 -102.42 0.36 [MPa3] cMF m 3.78 0.18 1.58 0.46 [%F 0] iQL m 2.0 0.73 4.0 0.21 [%F 0] cQL m 3.38 0.18 3.14 0.27 [%F 0] cPS m -0.10 0.76 0.62 0.18 [%F 0] iRA m -10.08 0.013 -14.64 0.012 [%F 0] cRA m -11.10 0.027 -14.10 0.027 [%F 0] iIO m -6.06 0.023 -14.04 0.003 [%F 0] cIO m -5.58 0.009 -10.44 0.007 [%F 0] iES m -3.4 0.14 -3.2 0.26 [%F 0] cES m -3.82 0.17 -2.19 0.67 [%F 0] iEO m -3.80 0.005 -3.4 0.033 [%F 0] cEO m -0.004 0.99 -2.2 0.31 [%F 0] Abbreviations: ipsilateral side (i), contralateral side (c), multifidus (MF), quadratus lumborum (QL), psoas (PS), rectus abdominis (RA), internal oblique (IO), external oblique (EO) and erector spinae (ES). Significant effects (p<0.05) are bolded. All muscle m forces are normalized to the muscle group’s maximum isometric force property (F 0).

110

Figure 5.3. Individual changes in the contralateral multifidus muscle force production over the stance phase of running with simulated core muscle fatigue in the CS group during the pre-CSKP kinematic condition. Muscle force is normalized for each m participant to their muscle group’s maximum isometric force (F 0).

111

Figure 5.4. Individual changes in peak force production of the deep core muscles in the CS group acting as compensators in response to a subject-specific level of overall core muscle fatigue. Compensations are shown for each participant in both their ‘pre’ and ‘post’ CSKP kinematic condition. Muscle forces are normalized to the muscle group’s m maximum isometric force property (F 0).

112

5.5 Discussion

Using kinematically-driven simulations, this study found that simulated core muscle fatigue during running and the adoption of a neuromuscular compensation strategy in response to that fatigue was associated with decreased compressive spinal loading and superficial core muscle force production and increased deep core muscle force production. These relationships were not significantly affected by a change in running kinematics, suggesting the consequences of core muscle fatigue and an adopted neuromuscular compensation strategy may be similar regardless of a chosen kinematic state.

Similar results were found in the group of runners that experimentally experienced a meaningful reduction in core stability following a core muscle fatiguing protocol (CS group) and a group that was able to maintain their core stability in the presence of muscle fatigue (NCS group). This study isolated the effect of simulated muscle fatigue on running biomechanical variables and fatigue was simulated the same way for both groups, leading to similar neuromuscular compensations..

An increase in deep core muscle force production was found to be the primary neuromuscular compensation utilized to maintain running kinematics in the presence of simulated core muscle fatigue. These deep core muscles were also fatigued in the simulations, however they were still able to increase their peak force production during running. This phenomenon may have occurred since fatigue was simulated as a reduction

m in the maximum isometric force property (F 0) of individual muscles, and during

m running, the deep core muscles produced forces much lower than their F 0. Previous

113 research supports this finding, as low levels of force production in the deep core musculature have been shown to be sufficient to stabilize the spine during upright dynamic tasks (Cholewicki and McGill 1996). Therefore, although these muscles were fatigued, they were still physically able to produce the required amount of force to maintain running kinematics in the presence of overall core muscle fatigue.

The results of this study suggest that if runners have the capability to produce sufficient force in the deep core muscles, they may be able to utilize a neuromuscular compensation strategy when the core is compromised and avoid adopting potentially detrimental movement patterns. Clinically, weakness of the deep core muscles, likely resulting from inhibition and disuse of these muscles (Hodges, Holm et al. 2006;

Wallwork, Stanton et al. 2009), has been frequently associated with low back disorders

(Freeman, Woodham et al. 2010). Training programs focusing on improving deep core muscle strength and activation have been associated with an increase in spinal stabilization and a reduction in low back pain symptoms (Sung 2003; Hides, Stanton et al. 2008; Freeman, Woodham et al. 2010). The findings from this study further suggest the influential role the deep core muscles may play in running.

Utilizing a neuromuscular compensation strategy was not significantly associated with any changes in lower extremity loading, suggesting that adopting this strategy may not influence a runner’s lower extremity injury risk. This strategy also required no significant change in estimated total energy consumption during running. This finding suggests that a neuromuscular compensation strategy in response to core muscle fatigue during running may not be as energetically costly as previously reported neuromuscular

114 compensation strategies, which have been shown to generally be energetically inefficient when adopted in response to lower extremity muscle weakness during walking (van der

Krogt, Delp et al. 2012). Since energy cost is believed to be a driving factor in running mechanics (Cavanagh and Williams 1982; Alexander 1984; Sparrow 2000), these results support the feasibility of this type of strategy. Once runners have been trained to engage their deep core muscles and increase their force production when necessary, they may be able to maintain this strategy throughout training regimens and over time.

In this study, fatigue was simulated as a reduction in a muscle’s capacity to

m produce force (decreased F 0). Since the ability of the trunk musculature to produce and maintain sufficient force during dynamic tasks is a key component of core stability

(Granata, Slota et al. 2004), simulating isolated trunk muscle fatigue may recreate a condition of poor core stability (Hart, Kerrigan et al. 2009). Therefore, utilization of a neuromuscular compensation strategy may be beneficial for runners lacking sufficient core stability. In previous work, we found that in response to experimentally reduced core stability, a group of runners naturally adopted ‘higher-risk’ running kinematics associated with increased lower extremity loading (Table 4.2). Results from this study suggest that training runners to engage and increase force production of the deep core musculature may facilitate the utilization of potentially ‘lower-risk’ neuromuscular compensation strategies when core stability is compromised. Future work may consider repeating these methods with additional components of core stability affected, such as activation timing or activation rate, and comparing the results to this study.

115

5.5.1 Limitations

There are some limitations that must be considered when interpreting the results of this study. Muscle forces in the model are calculated using Static Optimization (SO) in

OpenSim, which resolves the net joint moments into individual muscle forces at each instant in time by minimizing a performance criterion (sum of squared muscle activations) (Hicks and Dembia 2014). SO does not integrate forward in time and as a result, does not account for any anticipatory muscle activations that may occur in preparation for future movements or loading. Since the trunk muscles have been shown to activate prior to any voluntary lower extremity movements (Hodges and Richardson

1997), it is possible the simulated muscle activations and muscle forces in this study do not fully characterize the trunk muscle control strategies that may be used in vivo.

However since the body experiences the greatest external loading during the stance phase of running, the core muscles activate and play a large role during this time in the development and control of forces and loads at the joints. Therefore, the results of this study likely reflect the majority of the primary changes in control strategies that would occur with core muscle fatigue.

Energy consumption in this study was estimated as the sum of total cubed muscle stress (Crowninshield and Brand 1981). This estimate is dependent on simulated muscle force production. For the purposes of this study, this estimation was sufficient to begin to explore how energy consumption during running may be affected by muscle fatigue and a neuromuscular compensation strategy. Future work specifically interested in energy cost

116 during running may consider using forward dynamic simulations, allowing for more complex models of energy expenditure to be utilized (Umberger, Gerritsen et al. 2003).

Additionally, the small sample size in this study is a limitation that must be acknowledged. Figures 5.3 and 5.4 highlight some of the individual variability in the data which may have prevented significant differences from being detected. While within- subject analyses allow a smaller sample size to be utilized, a larger group of participants may have allowed for further characterization and stronger identification of the different neuromuscular compensation strategies that may be adopted in the presence of muscle fatigue.

Lastly, simulating the reduced capacity of a muscle to produce force is only one component of muscle fatigue. Muscular fatigue has been shown to affect not only force production, but also muscular response times, rate of force development, and force accuracy and steadiness (Kearney and Stull 1981; Parnianpour, Nordin et al. 1988;

Paasuke, Ereline et al. 1999; Singh, Arampatzis et al. 2010). Future work further interested in investigating the consequences of core muscle fatigue and subsequently adopted neuromuscular compensation strategies should consider simulating fatigue by limiting activation of muscles rather than force production and compare the results to this work. Additionally, using a more complex simulation workflow would allow for the manipulation of other components of fatigue such as muscle activation timing or activation rate.

117

5.6 Conclusion

In conclusion, core muscle fatigue and the subsequently adopted neuromuscular compensation strategy during running were associated with decreased spinal loading and superficial core muscle force production and increased deep core muscle force production during the stance phase of running. This compensation strategy may be favorable over ones where movement patterns are altered, as it was not associated with any change in estimated energy consumption or lower extremity loading during the stance phase of running. Only an increase in force production of the deep core muscles was required to execute this compensation strategy, suggesting that training runners to engage the deep core musculature in the presence of core muscle fatigue and/or poor core stability may allow them to utilize potentially lower-risk compensation strategies.

118

Chapter 6 : Research Findings and Future Work

6.1 Contributions

The main purpose of this research was to establish the effect of core stability on running-related injury mechanisms that have previously been identified to influence running injury risk. A more comprehensive understanding of how core stability affects running mechanics and the identification of potential compensation strategies used for poor core stability may ultimately contribute to the development of more effective and robust running injury prevention and rehabilitation regimens. First, a novel core stability knockdown protocol (CSKP) was utilized to experimentally create a quantified deficit in core stability in novice runners and identify associated changes in lower extremity joint angles and external joint moments during running. Then, a more physiologically accurate

OpenSim full-body model was developed and validated in order to facilitate the use of dynamic simulations to investigate full-body dynamics and contributions of the trunk muscles to dynamic tasks. Next, the experimentally collected running kinematic and kinetic data was used in combination with the newly developed OpenSim full-body model to create subject-specific simulations of novice runners jogging. These simulations enabled the isolation and investigation of different compensation strategies that may be utilized in response to reduced core stability. This innovative approach using experimental techniques in conjunction with subject-specific simulations to provide 119 clinical insight into running-related injury mechanisms in novice runners has not been done before. The major contributions of this work are listed below.

An experimentally-induced temporary decrease in core stability in novice runners was associated with an increased external peak knee flexion moment (pKF) during the stance phase of running (Chapter 2).

Novice runners with reduced core stability following a fatiguing core exercise protocol experienced a significant increase in the external pKF moment (13.5±2.5

%BW*h vs 14.3±3.1 %BW*h, p=0.001) during the stance phase of running. It has been theorized that insufficient core stability may lead to less efficient movements and ultimately musculoskeletal injury (Fredericson and Moore 2005; Willson, Dougherty et al. 2005; Kibler, Press et al. 2006). The observation in this study that reduced core stability was associated with an increase in the pKF moment is consistent with this theory. An increase in this moment has previously been associated with increased patellofemoral contact pressure during running (Andriacchi, Kramer et al. 1985; James

1995), which has been linked to patellofemoral pain syndrome (PFPS), one of the most common running injuries (Heino Brechter and Powers 2002). Therefore, insufficient core stability in novice runners may be a risk factor for PFPS.

120

An OpenSim full-body lumbar spine (FBLS) musculoskeletal model was developed and the model’s properties and jogging simulation results compared well to experimental data

(Chapter 3).

A more physiologically accurate OpenSim full-body model was developed and validated. The ability of a biomechanical simulation to produce results that can translate to real-life situations is largely dependent on the physiological accuracy of the musculoskeletal model. Very few full-body models were previously available in

OpenSim, and those that did exist were very limited in terms of trunk musculature and degrees of freedom in the spine (Hamner, Seth et al. 2010; Caruthers, Thompson et al.

2013). Building upon three previously developed OpenSim models (Arnold, Ward et al.

2010; Christophy 2010; Hamner, Seth et al. 2010), the FBLS model was developed and includes degrees of freedom in the lumbar spine as well as a detailed description of the eight major muscle groups of the trunk. The FBLS model’s trunk muscle geometry, maximal isometric joint moments, and muscle activations simulated during jogging were found to compare well to experimental data. The FBLS model was made freely available

(https://simtk.org/home/fullbodylumbar) for others to perform additional analyses and develop simulations investigating full-body dynamics and contributions of the trunk muscles to dynamic tasks.

121

A kinematic compensation strategy adopted by novice runners in response to experimentally reduced core stability was associated with increased knee loading during the stance phase of running (Chapter 4).

This is the first study, to the our knowledge, that specifically investigated how individuals lacking sufficient core stability may compensate for this deficiency during running and identified the associated biomechanical consequences. Using kinematically- driven simulations, it was found that altering movement patterns in response to reduced core stability (kinematic compensation strategy) was associated with an increased peak patellofemoral joint reaction force (6.2±1.0 BW vs 6.6±0.7 BW, p=0.029), peak knee abduction moment (4.3±0.9 %BW*h vs 5.1±0.8 %BW*h, p=0.01), knee abduction impulse (0.6±0.2 %BW*h*s vs 0.7±0.3 %BW*h*s, p=0.02), and peak knee extension moment (16.4±1.5 %BW*h vs 17.2±2.4 %BW*h, p=0.09) during the stance phase of running (all moments and impulses in this study were internal). Increases in these loading variables have previously been associated with patellofemoral pain and iliotibial band syndrome, two of the most common running injuries (Andriacchi, Kramer et al. 1985;

James 1995; Heino Brechter and Powers 2002; Fairclough, Hayashi et al. 2006;

Stefanyshyn, Stergiou et al. 2006; Powers 2010). These results suggest that novice runners with insufficient core stability may adopt altered movement patterns associated with increased lower extremity loading and ultimately a higher risk of sustaining a running-related injury.

122

A reduction in estimated energy consumption, spinal loading, and the workload of the core musculature may be ultimately driving the adoption of a kinematic compensation strategy in the presence of reduced core stability (Chapter 4).

Using kinematically-driven simulations, it was also found that adopting a kinematic compensation strategy in response to reduced core stability was associated with a reduction in estimated energy consumption (3249.3±1317.5 MPa3 vs

3009.9±1259.1 MPa3, p=0.059), cumulative lumbar spinal compression (p≤0.06), and reduced core muscle force production (p≤0.06) during the stance phase of running. Since running mechanics are believed to be largely driven by energy cost (Cavanagh and

Williams 1982; Alexander 1984; Sparrow 2000), the reduction in estimated energy consumption, along with a reduction in spinal loading and core muscle workload, may be the factors ultimately driving novice runners to adopt a kinematic compensation strategy when core stability is compromised.

A neuromuscular compensation strategy for core muscle fatigue was not associated with any change in estimated energy consumption or lower extremity loading during the stance phase of running (Chapter 5).

A neuromuscular compensation strategy, in which a runner alters only muscular activation strategies rather than movement patterns, may be one type of strategy utilized by runners to compensate for core muscle fatigue and/or a deficiency in core stability.

However, it has been suggested that neuromuscular compensation strategies may be energetically costly, so their feasibility has been questioned (van der Krogt, Delp et al.

123

2012). Using kinematically-driven simulations, it was found that utilizing a neuromuscular compensation strategy during running in the response to core muscle fatigue was associated with decreased lumbar spine compression (p≤0.08), decreased superficial core muscle force production (p≤0.033), and no change in lower extremity loading or energy consumption (p>0.1). These results suggest that this type of compensation strategy may not affect a runner’s lower extremity injury risk or energy cost during running. Therefore, a neuromuscular strategy may be a ‘low-risk’ strategy that is feasible for runners to learn and maintain, as a significant change in energy consumption is not required.

An increase in force production of the deep core musculature was the only muscular compensation required to utilize a neuromuscular compensation strategy during running in the presence of core muscle fatigue (Chapter 5).

Clinically, the deep core muscles are believed to play a large role in low back injury prevention and rehabilitation. Improving function of the deep core muscles has been associated with increased in spinal stabilization and a reduction in low back pain symptoms (Sung 2003; Hides, Stanton et al. 2008; Freeman, Woodham et al. 2010).

Although between-participant variability prevented significant muscular compensations from being detected, it was observed that the only apparent compensation required to maintain running kinematics in the presence of overall core muscle fatigue was an increase in deep core muscle force production. These results highlight the potential importance of these muscles in runners and suggest that by increasing engagement and

124 force production of the deep core muscles when the core is compromised, a runner may be able to adopt a ‘lower-risk’ neuromuscular compensation strategy rather than potentially detrimental movement patterns when the core muscles are fatigued and core stability is compromised.

A neuromuscular compensation strategy for reduced core stability during running may be favorable over a kinematic compensation strategy, as it prevents runners from adopting potentially detrimental running kinematics associated with increased lower extremity loading and injury risk (Chapters 4 and 5).

It is unknown how individuals lacking sufficient core stability may compensate for this deficiency during running. Identification of compensation strategies that may be utilized in the presence of reduced core stability and their biomechanical consequences provides insight into what injuries may arise in runners with this deficiency and why certain compensations are adopted over others. Two different types of compensation strategies were investigated. It was found that a kinematic compensation strategy adopted in response to reduced core stability was associated with increased knee loading during the stance phase of running that has previously been associated with increased running- related injury risk. In contrast, a neuromuscular compensation strategy had no detrimental consequences related to lower extremity injury risk. Therefore, although runners may prefer to adopt a kinematic compensation strategy when the core is compromised since it may require less energy consumption, the results from this study suggest utilizing a

125 neuromuscular compensation strategy may be more favorable in regards to a runner’s lower extremity injury risk.

6.2 Future Work

The work presented in this dissertation has added to the knowledge of how core stability affects running mechanics in novice runners and what potential compensation strategies may be used by runners with poor core stability. However, in order to use core stability to develop more effective and robust running injury prevention and rehabilitation regimens, more research needs to be performed. The following topics can build off the current research to further explore the impact of core stability on running-related injury incidence.

Additional runner and athlete populations

This research investigated the effect of core stability on running mechanics in novice runners, since novice runners are believed to be the most susceptible to developing running injuries (Buist, Bredeweg et al. 2010; Tonoli, Cumps et al. 2010;

Verhagen 2012; Schmitz, Russo et al. 2014). It is possible that core stability may affect more experienced runners and other athletes differently. In order to fully understand how core stability affects populations other than novice runners, it is important repeat similar analyses with these other populations of interest. This work would assist in the development of optimal training and injury prevention programs for specific types of athletes.

126

Effects of a core stability exercise intervention on running biomechanics

Chapters 2 and 4 showed that a reduction in core stability was associated with increased lower extremity loading variables that have previously been associated with increased running injury risk. With this evidence that core stability has the ability to affect these biomechanical variables, the next step is to design a core stability exercise intervention and determine whether interventions aimed at improving core stability can alter the same biomechanical variables investigated in this study to be ‘lower risk.’ This work would provide even further evidence regarding the clinical utility of core stability programs.

Develop a core neuromuscular training program specifically for runners

The core exercises included in the core stability knockdown protocol in Chapter 2 consisted of 4 dynamic and 4 isometric exercises chosen by our research staff to target both the superficial and deep core musculature with minimal involvement from the lower extremity muscles. When choosing these exercises, other considerations included how various exercises would affect passive motion capture markers and electromyography electrodes concurrently worn during the protocol. Therefore, while the exercises included in this protocol certainly provide a template for future training programs and the data from Chapter 5 highlight the importance of including the deep core musculature in training, it is not known if the core exercises included in this research are necessarily the optimal exercises for running injury prevention. It is necessary to develop and compare

127 multiple core neuromuscular training programs to determine which exercises are the most effective and efficient at reducing lower extremity loading.

Prospective study of core stability and running injury incidence

Chapters 2 and 4 showed that a reduction in core stability was associated with increases in knee loading variables during the stance phase of running that have been previously associated with increased PFPS and ITBS injury risk. A study prospectively measuring core stability and running injury incidence is required to identify if poor core stability is a causative factor in running injury incidence. This study may also lead to the identification of a threshold of core stability below which a runner is at increased injury risk. Similarly, comparing the injury incidence in a group of runners participating in a core stability training program with a group not participating in a program would also provide direct insight into the effect of core stability training programs on running-related injury incidence.

Further improvement of the FBLS musculoskeletal model

The FBLS model developed in Chapter 3 is the first OpenSim model to our knowledge to combine a complex model of the trunk, involving detailed trunk musculature and degrees of freedom in the lumbar spine, with a well-established

OpenSim lower extremity model. While this model is a vast improvement over the previously available full-body OpenSim models, there are still additional components that need to be incorporated into the model to make it more physiologically accurate.

128

Intra-abdominal pressure (IAP) and the transversus abdominis (TrA) muscle are both believed to contribute to trunk and spinal stability (Cholewicki, Juluru et al. 1999;

Hodges 1999) and would be beneficial additions to the FBLS model in order to further investigate their roles in running.

Incorporate additional components of core stability into simulations

Many components contribute to one’s ability to stabilize their core. These include core muscle strength, endurance, and neuromuscular coordination. Fatigue of the core musculature may negatively affect one or more of these components of core stability and lead to a reduced ability to control the trunk (Parnianpour, Nordin et al. 1988; Potvin and

O'Brien 1998; Granata and Gottipati 2008; Downey, Kamalapurkar et al. 2016). In

Chapter 5, we simulated core muscle fatigue as a surrogate for reduced core stability by decreasing the capacity of the core muscles to produce force. This is only one aspect of muscle fatigue that may affect core stability, so future work may consider exploring the effect of altering additional components that may be related to core stability, such as activation timing or activation rate, and compare the results to those found in Chapter 5.

6.3 Summary

Running is a physical activity with many health benefits that continues to grow in popularity each year. However, the annual running injury rate has been reported to be as high as 74% (Marti, Vader et al. 1988; Macera, Powell et al. 1989; van Mechelen 1992;

Buist, Bredeweg et al. 2010; Daoud, Geissler et al. 2012) and novice runners may be at

129 the highest risk of developing these injuries (Buist, Bredeweg et al. 2010; Tonoli, Cumps et al. 2010; Verhagen 2012; Schmitz, Russo et al. 2014). Previous research has shown core stability may affect lower extremity function (Bouisset 1991; Hodges and

Richardson 1997), which has led to the popular notion that insufficient core stabilization may lead to movements that are less efficient and ultimately contribute to musculoskeletal injury (Willson, Dougherty et al. 2005; Zazulak, Hewett et al. 2007).

However, the role that core stability plays during running and its influence on injury risk in runners is not well understood. In Chapter 2 of this dissertation, we experimentally investigated the direct downstream effects of reduced core stability in novice runners on running mechanics that have previously been associated with running injuries using a fatiguing core stability knockdown protocol (CSKP). Reduced core stability was significantly associated with an increased peak knee flexion moment during the stance phase of running, which has previously been associated with increased patellofemoral contact pressure during running. Therefore, insufficient core stability in novice runners may be a risk factor for development of patellofemoral pain syndrome.

It is not well understood how individual runners compensate for poor core stability and the consequences of those compensations. This knowledge may lead to the identification of runners lacking core stability, provide insight into what injuries may arise in runners with this deficiency, and ultimately contribute to the development of improved injury prevention and rehabilitation protocols. In Chapters 4 and 5, the experimentally collected data from Chapter 2 was used as input into OpenSim in conjunction with the novel musculoskeletal model developed in Chapter 3 to further

130 isolate and investigate the consequences of utilizing different possible compensation strategies for reduced core stability. Altering movement patterns to compensate for poor core stability (kinematic compensation) may be one strategy commonly used by runners when core stability is compromised (Rudolph, Axe et al. 2001; Monaghan, Delahunt et al. 2006). Using subject-specific simulations, it was found that a kinematic compensation strategy adopted in response to reduced core stability was associated with increased internal knee loading during the stance phase of running that has previously been associated with increased running injury risk (Dissertation Chapter 4). This strategy was also associated with a reduction in energy consumption, spinal loading, and the workload of the core musculature, which may be factors ultimately driving the adoption of this compensation strategy.

A neuromuscular compensation strategy, in which a runner alters only muscular activation strategies rather than movement patterns, may be another type of strategy utilized by runners to compensate for a deficiency in core stability. In Chapter 5, we investigated the effect of core muscle fatigue and a subsequently adopted neuromuscular compensation strategy on running mechanics and found an increase in deep core muscle force production was the only muscular compensation required to achieve this strategy and it was not associated with any change in estimated energy consumption or lower extremity loading during the stance phase of running. Therefore, neuromuscular training with an emphasis on deep core muscle engagement and force production may be beneficial for runners, potentially enabling them to utilize lower-risk compensation

131 strategies in regards to lower extremity injury, such as a neuromuscular strategy rather than a kinematic strategy, when core stability is compromised.

This is the first study to use experimental techniques in combination with subject- specific simulations to provide clinical insight into the relationship between core stability and running-related injury mechanisms in novice runners. We were able to demonstrate the impact of reduced core stability on biomechanical variables previously associated with running-related injuries and identify and compare the consequences of different compensation strategies that may be used by runners lacking sufficient core stability.

This work will lead to future research investigating the efficacy of interventions aimed at improving core stability in ultimately reducing running injury incidence.

132

Appendix A : Supplemental Material for Chapter 2

For further exploratory analysis, mixed models were also created for each running parameter separately for each leg. Table A.1 shows the mixed-model results for the dominant leg, non-dominant leg, and combined legs separately. It is important to consider the reduction in sa mple size with individual legs.

The results in Table A.1 suggest that the core stability knockdown protocol

(CSKP) may affect the dominant and non-dominant limbs differently. An increased peak knee flexion moment (Andriacchi, Kramer et al. 1985; James 1995), peak knee abduction angle (Huberti and Hayes 1984; Mizuno, Kumagai et al. 2001; Powers 2003), and peak hip internal rotation angle (Lee, Anzel et al. 1994; Powers 2003) are changes that have been associated with increased running-related injury risk while decreased knee and hip adduction impulses (MacMahon, Chaudhari et al. 2000; Stefanyshyn, Stergiou et al.

2006) have been associated with lower injury risk and may be considered ‘protective’ actions. It seems from these results that reduced core stability may be associated more with detrimental changes in the non-dominant limb and protective changes in the dominant limb. Investigating this asymmetry is an area that future studies could explore to better understand the effect of reduced core stability on running mechanics and ultimately running injury risk

133

Table A.1. Change from pre- to post-CSKP and mixed model effects (slopes) for each of the biomechanical parameters. Results are expressed as the mean ± standard error (p-value). Analyses are shown for legs individually and together. The reported p-value indicates the significance of the CSKP effect. The best mixed model for some of the running parameters also included an interaction effect between CSKP and leg is also included with its respective p-value, if applicable. The effects reported in the table include data from both legs. The CSKP effect characterizes the relationship between a reduction in core stability and the respective biomechanical variable using mixed effects models. All moments and impulses are external.

Vertical Knee Hip Vertical Knee Knee Knee Hip Hip Hip Average Knee Internal Internal Impact Flexion Adduction Adduction Adduction Adduction Adduction Loading Abduction Rotation Rotation Peak Moment Moment Impulse Angle Moment Impulse Rate Angle [°] Angle Angle [BW] [%BW*h] [%BW*h] [%BW*h*s] [°] [%BW*h] [%BW*h*s] [BW/s] [°] [°]

Change 0.011 ± -0.84 ± 0.61 ± -0.40 ± -0.51 ± -0.097 ± 0.05 ± -0.43 ± -0.02 ± -0.08 ± -1.98 ± Dominant 0.046 0.57 0.41 0.42 0.31 0.037 0.66 0.73 0.15 0.03 1.90 Leg (0.82) (0.18) (0.13) (0.37) (0.10) (0.03) (0.94) (0.57) (0.91) (0.01) (0.32) (post-pre) 134 Change Non- 0.001 ± -0.33 ± 1.02 ± -2.04 ± 0.10 ± 0.005 ± 1.65 ± 0.63 ± 0.04 ± 0.01 ± -2.09 ± Dominant 0.039 0.57 0.20 0.65 0.10 0.020 0.93 0.80 0.20 0.03 0.71 Leg (0.97) (0.57) (0.001) (0.01) (0.62) (0.80) (0.11) (0.45) (0.87) (0.82) (0.02) (post-pre) Change 0.006 ± -0.59 ± 0.81 ± -1.22 ± -2.04 ± -0.05 ± 0.85 ± 0.10 ± 0.01 ± -0.04 ± -2.04 ± Combined 0.029 0.40 0.20 0.39 0.10 0.02 0.58 0.54 0.13 0.02 0.99 Legs (0.84) (0.16) (0.001) (0.005) (0.191) (0.046) (0.16) (0.85) (0.94) (0.07) (0.053) (post-pre)

0.003 ± -0.29 ± 0.41 ± -0.61 ± -0 10 ± -0.02 ± 0.43 ± 0.05 ± 0.005 ± -0.02 ± -1.02 ± CSKP 0.015 0.20 0.10 0.19 0.10 0.01 0.29 0.27 0.061 0.01 0.49 effect (0.84) (0.16) (0.001) (0.005) (0.191) (0.046) (0.16) (0.85) (0.94) (0.07) (0.053) CSKP and -0.41 ± 0 10 ± 0.02 ± 0.02 ± Leg 0.19 0.10 0.01 0.01 Interaction (0.049) (0.08) (0.028) (0.04) Effect Significant effects (p<0.05) are bolded. All GRF loading variables were normalized to participant body weight (BW) and moments and impulses were normalized to BW and height (h). ). Peaks angles and moments were analyzed.

Appendix B : Supplemental Methods and Results for the Development and Validation of the Full-Body Lumbar Spine (FBLS) Model

B.1 Supplemental Methods

B.1.1 Model Development

The Full-body Lumbar Spine (FBLS) model was developed by combining three previously built OpenSim models (Arnold, Ward et al. 2010; Hamner, Seth et al. 2010;

Christophy, Faruk Senan et al. 2012). The Hamner full-body model has previously been validated (Arnold, Ward et al. 2010; Hamner, Seth et al. 2010) and was used as the base model for the FBLS model. Almost all respective body and muscle properties for the lower extremities, pelvis, and arms remained unchanged from the Hamner model to the

FBLS model. The only exceptions were as follows: 1) the patella from Arnold’s lower limb model (Arnold, Ward et al. 2010) was added to Hamner’s lower extremities and 2)

Hamner’s psoas major muscle, which attaches from the lumbar spine to the distal femur, was replaced with the psoas major from Christophy’s detailed model of the lumbar spine

(Christophy, Faruk Senan et al. 2012), however the distal attachments from Hamner’s model were employed in the FBLS model. Christophy’s torso, with added head and neck geometry positioned to match those of the Hamner model, as well as the individual lumbar vertebrae bodies from Christophy’s model was used in place of Hamner’s torso in

135 the FBLS model. Christophy’s ribcage was also replaced by one with slightly different geometry that also included the left and right scapula.

First, all models were scaled to the same size before being combined. Next, the models were combined as described above and the resulting unscaled model had geometry representative of a 180 cm tall male. Since Christophy’s torso body did not include the head, neck, or scapulae the mass and inertial properties of the FBLS model’s torso body (head, neck, thoracic spine, ribcage, scapulae) needed to be recalculated. To determine the new model’s torso mass, the mass of the lumbar vertebrae were subtracted from Hamner’s torso mass. Hamner’s torso inertial properties were scaled to this new mass and these were used as the new inertial properties for the torso of the FBLS model.

The model’s torso mass center was then moved to match that of Hamner’s model and the parallel axis theorem was used to recalculate the torso inertial properties about this new mass center. It is important to note in a generic model, the torso body’s mass center and inertial properties are essentially an ‘initial guess’ of the true mass properties. Since the torso is the largest body with the most uncertain mass properties, when inverse dynamics is performed in OpenSim using the residual reduction algorithm (RRA), the mass properties (mass center and inertial properties) of the torso are systematically adjusted to achieve a dynamically consistent set of kinematics and kinetics (Delp, Anderson et al.

2007).

After scaling the component models, discrepancies still remained since Hamner’s model and Christophy’s model used different ribcage and pelvis geometry. Consequently, all attachment points of the 224 trunk muscle fascicles to the ribcage and pelvis had to be

136 adjusted to their appropriate physiological locations in the FBLS model. In order to reattach these muscles, the FBLS model and Christophy’s lumbar spine model were closely inspected side by side. Each individual muscle fascicle in the FBLS model that was not attached in the proper place due to geometry discrepancies was carefully manually adjusted to attach to the anatomically correct position, as determined by

Christophy in his lumbar spine model.

B.1.2 Altered Maximum Isometric Forces

m We made changes to the maximum isometric force property (F 0) of some of the

m trunk and lower extremity muscles as shown in Table B.1. When comparing the F 0 of some of the trunk muscle groups in Christophy’s lumbar spine model to those in

Hamner’s full-body model, we found some discrepancies. In order to be consistent with

m the base model, we increased F 0 for the internal oblique, external oblique, and psoas major muscle fascicles by a scale factor equal to the ratio describing the strength of the respective muscle in Hamner’s model to the total strength of the corresponding muscle group in Christophy’s model. Additionally, upon comparison to a cadaveric study done

m by Delp (Delp, Suryanarayanan et al. 2001), we decided to reduce F 0 for the rectus abdominis. Lastly, when performing simulations of running, we found some of the lower extremity muscles were too weak for the simulation to run successfully and also scaled

m up their F 0.

137

m m Muscle Original F 0 New F 0 Scaling Factor Gmed1 819.0 1288.5 1.5 Gmed2 573.0 859.5 1.5 Gmed3 653.0 979.5 1.5 Gmin1 270.0 405.0 1.5 Gmin2 285.0 427.0 1.5 Gmin3 323.0 484.0 1.5 Soleus 3549.0 5323.5 1.5 MedGas 1558.0 2337.0 1.5 LatGas 683.0 1024.5 1.5 FlexHal 322.0 483.0 1.5 FlexDig 310.0 465.0 1.5 PerLong 943.0 1414.5 1.5 TibPost 1588.0 2382.0 1.5 Iliacus 1073.0 1609.5 1.5 Tfl 233.0 349.5 1.5 Semitend 410.0 615.0 1.5 IL_L1 50.0 97.3 1.9 IL_L2 71.0 138.1 1.9 IL_L3 84.0 163.4 1.9 IL_L4 87.0 169.2 1.9 IL_R5 11.0 21.4 1.9 IL_R6 14.0 27.2 1.9 IL_R7 18.0 35.0 1.9 IL_R8 16.0 31.1 1.9 IL_R9 23.0 44.7 1.9 IL_R10 46.0 89.5 1.9 IL_R11 57.0 110.9 1.9 IL_R12 68.0 132.3 1.9 LTpL_L1 36.0 70.0 1.9 m Table B.1. Changes made to the maximum isometric force property (F 0) in the FBLS model for some lower extremity and trunk muscles and the scaling factor that was applied. Muscle fascicle abbreviations correspond to the name of the fascicle in OpenSim (continued).

138

Table B.1 continued

m m Muscle Original F 0 New F 0 Scaling Factor LTpL_L2 42.0 81.7 1.9 LTpL_L3 47.0 91.4 1.9 LTpL_L4 51.0 99.2 1.9 LTpL_L5 53.0 103.1 1.9 LTpT_T1 13.0 25.3 1.9 LTpT_T2 26.0 50.6 1.9 LTpT_T3 26.0 50.6 1.9 LTpT_T4 10.0 19.4 1.9 LTpT_T5 10.0 19.4 1.9 LTpT_T6 15.0 29.2 1.9 LTpT_T7 18.0 35.0 1.9 LTpT_T8 29.0 56.4 1.9 LTpT_T9 34.0 66.1 1.9 LTpT_T10 37.0 72.0 1.9 LTpT_T11 38.0 73.9 1.9 LTpT_T12 32.0 62.3 1.9 LTpT_R4 10.0 19.4 1.9 LTpT_R5 10.0 19.4 1.9 LTpT_R6 15.0 29.2 1.9 LTpT_R7 18.0 35.0 1.9 LTpT_R8 29.0 56.4 1.9 LTpT_R9 34.0 66.1 1.9 LTpT_R10 37.0 72.0 1.9 LTpT_R11 38.0 73.9 1.9 LTpT_R12 32.0 62.3 1.9 EO1 90.0 111.7 1.2 EO2 106.6 132.4 1.2 EO3 111.9 138.9 1.2 EO4 108 134.1 1.2 Continued

139

Table B.1 continued

m m Muscle Original F 0 New F 0 Scaling Factor EO5 125.6 156.0 1.2 EO6 182.8 227.0 1.2 IO1 85.2 124.0 1.4 IO2 103.2 150.1 1.4 IO3 104.0 151.2 1.4 IO4 123.1 179.0 1.4 IO5 108.0 157.1 1.4 IO6 95.3 138.6 1.4 Ps L1L2 IVD 55.2 91.3 1.6 Ps L2L3 IVD 54.7 90.6 1.6 Ps L3L4 IVD 16.6 27.4 1.6 Ps L4L5 IVD 36.3 60.1 1.6 Ps L1 TP 28.0 46.3 1.6 Ps L2 TP 970 160.5 1.6 Ps L3 TP 46.5 76.9 1.6 Ps L4 TP 74.0 122.4 1.6 Ps L5 TP 79.6 131.7 1.6 Ps L1 VB 97.0 160.5 1.6 Ps L5 VB 87.9 145.4 1.6

B.2 Supplemental Results

B.2.1 Full Table of Trunk Muscle Moment Arms

Table B.2 shows all moment arms for individual trunk muscle fascicles and how they compare to moment arms measured experimentally by Jorgensen et al (Jorgensen,

Marras et al. 2001). If no symbol is associated with a fascicle’s calculated moment arm

140 this means there were no experimental data available for the fascicle. Experimental moment arms for the multifidus muscle were not reported in the literature.

Moment Arm (mm) with Respect to Given Lumbar Vertebral Joint Level Fascicle L1-2 L2-3 L3-4 L4-5 L5-S1 Multifidus

m1s 33 37 32 N/A N/A m1t1 41 45 39 25 N/A m1t2 41 47 45 35 15 m1t3 38 52 55 52 38 m2s N/A 41 37 26 N/A m2t1 N/A 47 44 33 14 m2t2-3 N/A 42 49 50 40 m3s N/A N/A 36 32 17 m3t1-3 N/A N/A 46 54 49 m4s N/A N/A N/A 37 34 m4t1-m4t3 N/A N/A N/A 46 43 m5s N/A N/A N/A N/A 28 m5t1-m5t3 N/A N/A N/A N/A 38 m1_laminar 30 32 N/A N/A N/A m2_laminar N/A 33 32 N/A N/A m3_laminar N/A N/A 36 30 N/A m4_laminar N/A N/A N/A 35 25 m5_laminar N/A N/A N/A N/A 35 AVG 37 42 41 38 32 Iliocostalis Lumborum

IL_L1 28 ^^ 43^ 51* 53* 43^^ IL_L2 N/A 33^^ 45^ 50** 43^^ IL_L3 N/A N/A 32^^ 36** 29^^ IL_L4 N/A N/A N/A 27^^ 24^^ Table B.2. Sagittal plane moment arms for all trunk muscle fascicles. A *,**,^, or ^^ signifies the model’s moment arm was within 1 standard deviation (SD), 1.01-1.5 SD, 1.51-2 SD, or 2.01+ SD of the experimentally collected data, respectively (Jorgensen 2001). Muscle fascicle abbreviations correspond to the name of the fascicle in OpenSim (continued).

141

Table B.2 continued

Moment Arm (mm) with Respect to Given Lumbar Vertebral Joint Level Fascicle L1-2 L2-3 L3-4 L4-5 L5-S1 IL_R5 40^ 47** 47** 38** 25^^ IL_R6 40^ 47** 47^^ 40^^ 25^^ IL_R7 39^ 47** 48** 41^^ 24^^ IL_R8 39^ 46** 48** 42^^ 26^^ IL_R9 38^ 46** 47^ 40^^ 23^^ IL_R10 38^ 45** 45^ 37^^ 19^^ IL_R11 38^ 44^ 43^^ 33^^ 13^^ IL_R12 28^^ 36 37^^ 30^^ 13^^ AVG 36.6^^ 43^ 45^ 39^^ 26^^ LongissimusThoracis

LTPL_L1 29^^ 43^ 51* 52** 41^^ LTPL_L2 N/A 32^^ 45^ 51** 46^^ LTPL_L3 N/A N/A 35^^ 45^^ 42^^ LTPL_L4 N/A N/A N/A 32^^ 32^^ LTPL_L5 N/A N/A N/A N/A 25^^ LTPT_T1 N/A N/A N/A N/A N/A LTPT_T2 40^ N/A N/A N/A N/A LTPT_T3 35^^ 46** N/A N/A N/A LTPT_T4 35^^ 48** 53* N/A N/A LTPT_T5 34^^ 44^ 50** N/A N/A LTPT_T6 34^^ 44^ 51* 50** N/A LTPT_T7 34^^ 43^ 47** 45^^ 30.75^^ LTPT_T8 35^^ 43^ 48** 51** 41^^ LTPT_T9 34^^ 42^ 48** 54* 51** LTPT_T10 33^^ 42^^ 47** 54* 51** LTPT_T11 35^^ 45^ 50** 53* 47^ LTPT_T12 34^^ 45^ 50** 51** 41^^ LTPT_R4 38^ 49* 54* N/A N/A LTPT_R5 35^^ 47** 53* N/A N/A LTPT_R6 36^^ 45^ 50** 50** N/A LTPT_R7 35^^ 41^^ 47** 46^^ 31^^ LTPT_R8 35^^ 41^^ 47** 50** 40^^ LTPT_R9 35^^ 41^^ 47** 53* 50^ LTPT_R10 35^^ 42^^ 47** 53* 50^ Continued

142

Table B.2 continued

Moment Arm (mm) with Respect to Given Lumbar Vertebral Joint Level Fascicle L1-2 L2-3 L3-4 L4-5 L5-S1 LTPT_R11 35^^ 42^^ 47** 52** 47^ LTPT_R12 33^^ 42^^ 47** 51** 42^^ AVG 35^^ 43^ 48** 50^ 42^^ Latissimus Dorsi

LD Il 9^ 19^ 28** 28^^ 20^^ LD L1 N/A N/A N/A N/A N/A LD L2 17** N/A N/A N/A N/A LD L3 17** 26** N/A N/A N/A LD L4 17** 28** 34^^ N/A N/A LD L5 18** 29** 36* 36^^ N/A LD T7 N/A N/A N/A N/A N/A LD T8 N/A N/A N/A N/A N/A LD T9 N/A N/A N/A N/A N/A LD T10 N/A N/A N/A N/A N/A LD T11 N/A N/A N/A N/A N/A LD T12 N/A N/A N/A N/A N/A LD R11 N/A N/A N/A N/A N/A LD R12 N/A N/A N/A N/A N/A AVG 15** 25** 33* 32^^ 20^^ Rectus Abdominis

RA 73^^ 76^^ 64^^ 64* 75* External Oblique

EO1 74* 85^^ 85^^ 90^^ 40* EO2 70* 80^^ 81^^ 87^^ 37* EO3 55** 67** 70^^ 80^^ 36* EO4 41^^ 54* 61^^ 73^^ 34* EO5 7^^ 10^^ 14* 25* 45* EO6 6^^ 4^^ 2^ 15* 36* AVG 42^^ 50* 52^ 62^^ 38* Internal Oblique

IO1 N/A N/A N/A N/A 37* IO2 N/A N/A N/A N/A 30** IO3 N/A N/A N/A N/A 36* IO4 57^^ 53^ 39* 31* 20^ Continued

143

Table B.2 continued

Moment Arm (mm) with Respect to Given Lumbar Vertebral Joint Level Fascicle L1-2 Fascicle L1-2 Fascicle L1-2 IO5 28^^ 22^^ 12^^ 8^ 16~ IO6 1^^ 5^^ 10^^ 7^ 6^^ AVG 29^^ 27^^ 20^ 16** 24^ Psoas

Ps L1L2 IVD 1^ 1** 2* 13^^ 32^^ Ps L2L3 IVD N/A 5* 8** 17^^ 36^^ Ps L3L4 IVD N/A N/A 4* 14^^ 33^^ Ps L4L5 IVD N/A N/A N/A 6** 26^^ Ps L1 TP 15** 15^ 10^ 11^^ 23^^ Ps L2 TP N/A 14^ 9* 3* 25^^ Ps L3 TP N/A N/A 15^^ 3* 20^^ Ps L4 TP N/A N/A N/A 4* 18^ Ps L5 TP N/A N/A N/A N/A 9* Ps L1 VB 2** 2* 2* 9^ 31 Ps L5 VB N/A N/A N/A N/A 15** AVG 6* 7* 7* 9^ 24^^ Quadratus Lumborum

QL ant I2 12_1 29* 35* 34* 25* 7^^ QL ant I2 T12 7 16^^ 23** 24** 12^^ QL ant I3 12_1 26** 30* 26* 15^^ 6^^ QL ant I3 12_2 28* 31* 27* 15^^ 6^^ QL ant I3 12_3 28* 31* 27* 15^^ 6^^ QL ant I3 T12 4^^ 12^^ 17^ 15^^ 3^^ QL mid L2 12_1 18^^ N/A N/A N/A N/A QL mid L3 12_1 23^ 25** N/A N/A N/A QL mid L3 12_2 24^ 24** N/A N/A N/A QL mid L3 12_3 24^ 24** N/A N/A N/A QL mid L3 12_3 26** 27* 21** N/A N/A QL post I1 L3 N/A N/A 28* 29* 20^^ QL post I2 L2 N/A 26* 29* 24** 9^^ QL post I2 L3 N/A N/A 24* 22** 10^^ QL post I2 L4 N/A N/A N/A 18^ 10^^ QL post I3 L1 22.75^ 27* 24* 15^^ 4^^ QL post I3 L2 N/A 24** 23** 15^^ 3^^ QL post I3 L3 N/A N/A 20^ 13^^ 2^^ AVG 21.6^^ 25* 25* 19^ 8^^

144

B.2.2 Spinal Compressive Loading

The FBLS model was further validated by comparing estimated axial compressive loads on each of the lumbar vertebrae to loads previous published in the literature. Since little to no data has been published on spinal compressive loads during running, Figure

B.1 shows spinal loads during standing. The loads estimated for the FBLS model represent the internal axial joint load acting on the given vertebra, calculated as a point load using OpenSim’s Joint Reaction analysis. Axial vertebral loads estimated using the

FBLS model fall within range of those previously published and the behavior of spinal loading also compares well. It is important to note, El-Rich et al and Callaghan et al both estimate these loads using different models developed in software packages making some different modeling assumptions than OpenSim (Callaghan and McGill 2001; El-Rich,

Shirazi-Adl et al. 2004). Most notably, these models account for the deformable, passive structures in the spine, whereas the FBLS model does not. El-Rich et al use a finite element model with deformable structures that represent the intervertebral (IV) discs and estimate the load on the disc superior to the given vertebrae in Fig. B.1, while Callaghan et al also account for passive tissue properties of the spine in their model. Nachemson et al measured pressure in vivo in the IV disc superior to the given vertebrae and used the measured pressure to calculate the total load experienced by the disc (Nachemson and

Morris 1964). The load experienced by the IV disc superior to a given vertebrae is similar but not identical to the internal load carried by the joint structure, as estimated in

OpenSim, acting on the given vertebra. Since the IV disc is not directly modeled in

145

OpenSim it is possible the FBLS model may slightly over estimate the magnitude of spinal loading.

Figure B.1. Estimated internal axial compressive loads on each of the lumbar vertebrae during standing compared to loads previous published in the literature (Nachemson and Morris 1964; Callaghan and McGill 2001; El-Rich, Shirazi-Adl et al. 2004). Reported compressive loads are normalized to body mass.

B.3 Simulated Muscle Forces

B.3.1 Standing

Relaxed standing with arms held up to the side was simulated for 1 subject

(female, 1.66 m, 59.05 kg) using the FBLS model in order to also validate the model

146 during a slow, simple task in addition to running. We compared trunk muscle forces simulated during standing to trunk muscle force predictions already published for this task in the literature (Table B.3). Since it is difficult to directly measure trunk muscle forces, it is important to keep in mind that the muscle forces predicted by the FBLS model are compared to forces estimated from other musculoskeletal models that include a variation of muscles and may make some different modeling assumptions than this study.

Also, data across these studies is collected from subjects with different anthropometry which may affect muscle force magnitude.

Table B.3 shows estimates for the total erector spinae (ES) muscle force during standing for the FBLS model compared to data previously published in the literature

(Schultz, Haderspeck et al. 1983; Calisse, Rohlmann et al. 1999; Wilke, Rohlmann et al.

2003; Rohlmann, Bauer et al. 2006). In the FBLS model and the model used by Schultz et al, the ES is separated into two muscle groups, the iliocostalis (IL) and the longissimus

(LT). The ES force reported in Table B.3 is the sum of the force estimate from the IL and

LT. The ES muscle force estimated using the FBLS model falls within the range of these other sources.

147

FBLS Schultz Rohlmann Wilke Calisse Model 1983 2006 2003 1999 RA 0 2 MF 16 42 LT 120 86 IL 87 82 LD 26 20 QL 3 10 PS 3 2 ES 207 168 170 100 250 (IL+LT) Table B.3. Comparison of compared trunk muscle forces (N) simulated during standing to trunk muscle force predictions previously published in the literature (Schultz, Haderspeck et al. 1983; Calisse, Rohlmann et al. 1999; Wilke, Rohlmann et al. 2003; Rohlmann, Bauer et al. 2006).

Table B.3 also compares muscle force estimates between the FBLS model and the model used by Schultz et al for the remaining trunk muscles including: the rectus abdominis (RA), multifidus (MF), latissimus dorsi (LD), quadratus lumborum (QL), psoas (PS), IL and LT. The reported muscle force is the total muscle force for the muscle group (left and right sides). These trunk muscle forces predicted during standing using the FBLS model also compare well to previously published data.

148

Figure B.2. Trunk and lower extremity muscle forces during jogging estimated using the FBLS model compared to those previously published using a more simplified full-body model (Hamner, Seth et al. 2010). Muscle forces are normalized to their maximum isometric force property. Muscle forces shown are all on the stance side, except the erector spinae (ES) which is the sum of both left and right sides.

B.3.2 Running

Figure B.2 shows simulated trunk and lower extremity muscle forces during jogging using the FBLS model compared to muscle forces during jogging previously published by Hamner et al using their full-body model (Hamner, Seth et al. 2010).

Muscle forces are normalized to their maximum isometric force property. Despite different musculoskeletal models and modeling assumptions used in each of these simulations, the timing of the estimated muscle forces compare quite well, however the

149 magnitude of the muscle forces are more variable. Arnold et al also published simulated lower extremity muscle forces during jogging (using a lower extremity musculoskeletal model only) and simulated forces using the FBLS model compare to these well also

(Arnold, Hamner et al. 2013). It is important to note, Hamner et al and Arnold et al used a slightly different optimization procedure to estimate muscle forces than used in this study. The previous studies used computed muscle control (CMC) in OpenSim, which performs integration to advance the system forward in time using a specified look-ahead window. It uses a proportional-derivative (PD) controller to track desired kinematics in combination with static optimization (SO) to distribute muscle forces across a joint.

Consequently, CMC allows for anticipatory activations, whereas SO used by itself (as in this study) is a frame-by-frame solver that resolves net joint moments into individual muscle forces at each instant in time (Thelen and Anderson 2006; Hicks and Dembia

2014). Additionally, Hamner et al constrained some of their muscle activation levels based on EMG data and no muscle activations constraints were used in this study. Lastly,

Hamner’s trunk muscles were extremely limited compared to the FBLS model and only included one muscle path (actuator) for each of the following muscle groups on each side: external obliques, internal obliques, and erector spinae muscle groups (Hamner,

Seth et al. 2010). The FBLS model included many more trunk muscle groups (not shown in Fig. B.2, see Fig 3.1) and represented all trunk muscle groups as multiple fascicles, accounting for the large surface area of these muscles. Since so few trunk muscles were included in Hamner’s 2010 model, it is likely that their model overestimated individual trunk muscle force production, and the muscle forces estimated by the FBLS model may

150 be more representative of those experienced during running. Besier et al also simulated lower extremity muscle forces during jogging (Besier, Fredericson et al. 2009) and the forces simulated using the FBLS model also compared well to their previously published data. The FBLS model and some example data have both been made freely available on the SimTK website (https://simtk.org/home/fullbodylumbar).

151

Appendix C : Full Linear Mixed Model Effects for Chapters 4 and 5

152

Table C.1. All results from linear mixed-models for estimated energy consumption and lower extremity loading variables. The group mean and standard deviation (SD) of each variable for each condition are listed along with the model effect and corresponding p-value. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be driven by a reduction in core stability. Variables defined as larger in the negative direction are indicated with (-). In the case where the best linear mixed model for a biomechanical parameter included only a kinematic effect, N/A is listed by the effects not included in the model. All forces, moments, and impulses are internal. Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are bolded and italicized. Forces are normalized to body weight (BW) and moments and impulses are normalized to BW*height (BW*h). Energy pFPFJ pTibComp (-) pKabdM (-) KAbdI (-) pKExtM [MPa^3] [BW] [BW] [%BW*h] [%BW*h*s] [%BW*h] CS group mean SD mean SD mean SD mean SD mean SD mean SD pre-none 3314.3 1387.5 6.2 1.0 3.5 0.5 -4.3 0.9 -0.6 0.2 16.4 1.5 pre-fatigue 3184.3 1453.9 6.2 1.0 3.5 0.5 -4.3 0.9 -0.6 0.2 16.4 1.5 post-none 3093.9 1327.0 6.6 0.7 3. 6 0.6 -5.1 0.8 -0.7 0.3 17.2 2.4 153 post-fatigue 2925.9 1385.3 6.6 0.7 3.8 0.6 -5.1 0.8 -0.7 0.3 17.2 2.4

kinematic effect -239.4 0.38 -0.30 -0.74 -0.10 0.76 p-value 0.059 0.029 0.17 0.01 0.02 0.09 fatigue effect -149.04 N/A N/A N/A N/A N/A p-value 0.21 N/A N/A N/A N/A N/A interaction effect -37.88 N/A N/A N/A N/A N/A p-value 0.87 N/A N/A N/A N/A N/A NCS group mean SD mean SD mean SD mean SD mean SD mean SD pre-none 1770.5 283.8 5.4 1.0 3.4 0.5 -5.0 1.3 -0.7 0.2 14.8 2.4 pre-fatigue 1682.0 506.3 5.4 1.0 3.4 0.5 -5.0 1.3 -0.7 0.2 14.8 2.4 post-none 1858.5 457.0 5.4 1.3 3.9 0.9 -4.3 1.8 -0.6 0.3 14.3 3.5 post-fatigue 1742.1 526.2 5.4 1.3 3.9 0.9 -4.3 1.8 -0.6 0.3 14.3 3.5 kinematic effect 74.06 -0.06 -0.44 0.68 0.10 -0.48 p-value 0.508 0.856 0.08 0.003 0.0005 0.58 fatigue effect -102.42 N/A N/A N/A N/A N/A p-value 0.36 N/A N/A N/A N/A N/A interaction effect -27.88 N/A N/A N/A N/A N/A p-value 0.9 N/A N/A N/A N/A N/A Abbreviations: peak patellofemoral joint reaction force (pFPFJ), peak tibial compression (pTibComp), peak knee abduction moment (pKAbdM), knee abduction impulse (KAbdI), peak knee extension moment (pKExtM).

Table C.2. Results from linear mixed-models for spinal compressive and anterior shear loading variables. The group mean and standard deviation of each variable for each condition are listed along with the model effect and corresponding p-value. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group). Variables defined as larger in the negative direction are indicated with (-). All loads reported on a given vertebra are those acting upon it from the superior vertebra. impS1 impL5 impL4 impL3 impL2 Comp Comp Comp Comp Comp [%BW*s] [%BW*s] [%BW*s] [%BW*s] [%BW*s] CS group mean SD mean SD mean SD mean SD mean SD pre-none 55.8 13.8 55.5 14.1 54.3 13.9 51.2 12.6 46.7 10.6 pre-fatigue 51.9 13.3 51.6 13.7 50.2 13.5 47.2 12.4 42.9 10.5 post-none 51.5 10.8 51.1 11.1 49.8 10.7 47.2 9.6 43.2 8.0 post-fatigue 47.7 11.5 47.3 11.9 46.0 11.6 43.3 10.6 39.5 9.0 kinematic effect -4.2 -4.4 -4.34 -4.0 -3.46 p-value 0.04 0.04 0.04 0.046 0.049 154 fatigue effect -3.8 -3.84 -3.94 -3.94 -3.72

p-value 0.06 0.06 0.06 0.047 0.038 interaction effect 0.12 0.12 0.16 0.17 0.004 p-value 0.97 0.97 0.96 0.96 0.99 NCS group mean SD mean SD mean SD mean SD mean SD pre-none 41.6 3.8 40.9 4.0 39.7 4.0 37.6 4.0 34.8 4.0 pre-fatigue 37.7 7.4 37.0 7.4 35.7 7.3 33.7 6.7 31.1 6.0 post-none 43.1 10.7 42.4 10.9 41.2 11.0 39.1 10.9 35.9 10.4 post-fatigue 37.8 9.3 37.1 9.5 36.1 9.8 34.1 9.6 31.4 9.1 kinematic effect 0.78 0.86 1.0 0.94 0.70 p-value 0.73 0.72 0.67 0.67 0.74 fatigue effect -4.4 -4.58 -4.54 -4.4 -4.1 p-value 0.08 0.08 0.08 0.076 0.076 interaction effect -1.36 -1.44 -1.12 -1.0 -0.88 p-value 0.77 0.77 0.81 0.82 0.83

Abbreviations: impulse of compressive load acting on S1 (impS1comp), peak compressive load acting on S1 (pS1comp), peak anterior shear load acting on S1 (pS1shear). All forces and impulses are internal. Significant effects (p<0.05) are bolded, trending effects (0.5≤p≤0.1) are bolded and italicized. All spinal loading variables are normalized to body weight (BW). Continued.

Table C.2 cont’d. pS1 pL5 pL4 pL3 pL2 impL1 [%BW*s] Comp Comp Comp Comp Comp Comp [%BW] [%BW] [%BW] [%BW] [%BW] CS group mean SD mean mean SD mean mean SD mean SD mean SD pre-none 43.9 9.7 386.8 381.4 134.2 357.8 381.4 134.2 357.8 122.2 321.5 103.6 pre-fatigue 40.2 9.5 356.1 350.4 131.7 326.5 350.4 131.7 326.5 119.5 293.1 100.0 post-none 40.7 7.2 381.7 380.3 139 358.5 380.3 139 358.5 130.1 325.3 109.9 post-fatigue 37.0 8.3 337.2 331.4 144.7 307.8 331.4 144.7 307.8 134.1 272.2 117.9 kinematic effect -3.2 -11.98 -10.1 -9.02 -10.1 -9.02 -8.6 p-value 0.06 0.47 0.55 0.57 0.55 0.57 0.53 fatigue effect -3.6 -37.6 -40.0 -41.0 -40.0 -41.0 -40.6 p-value 0.036 0.04 0.037 0.025 0.037 0.025 0.013 interaction effect -0.01 -13.76 -17.8 -19.4 -17.8 -19.4 -24.7 p-value 0.99 0.67 0.6 0.54 0.6 0.54 0.37 NCS group mean SD mean mean SD mean mean SD mean SD mean SD

155 pre-none 32.6 4.2 315.4 302.3 41.1 276.8 302.3 41.1 276.8 37.2 235.2 42.9 pre-fatigue 29.2 5.6 277.3 262.6 86.3 240.3 262.6 86.3 240.3 77.1 211.8 65.2

post-none 33.7 10.0 311.7 299.5 101.2 276.8 299.5 101.2 276.8 96.5 242.4 97.4 post-fatigue 29.4 8.8 278.8 269.1 88.0 249.3 269.1 88.0 249.3 85.1 223.9 80.2 kinematic effect 0.62 -1.1 1.84 4.44 1.84 4.44 9.68 p-value 0.75 0.97 0.95 0.88 0.95 0.88 0.71 fatigue effect -3.8 -35.4 -35.0 -31.8 -35.0 -31.8 -21.0 p-value 0.078 0.3 0.31 0.31 0.31 0.31 0.43 interaction effect -0.8 4.12 9.36 9.1 9.36 9.1 5.0 p-value 0.83 0.95 0.88 0.88 0.88 0.88 0.92 Continued

Table C.2 cont’d. pL1 pS1 (-) pL5 (-) pL4 (-) pL3 (-) pL2 (-) pL1 (-) Comp Shear Shear Shear Shear Shear Shear [%BW] [%BW] [%BW] [%BW] [%BW] [%BW] [%BW] CS group mean SD mean SD mean SD mean SD mean SD mean SD mean SD pre-none 301.1 91.3 16 4.6 17.1 4.7 33.5 10.4 68.0 27.7 92.8 36.7 88.1 33.7 pre-fatigue 272.0 88.8 16.1 4.2 17.5 3.5 28.0 7.0 59.0 22.2 81.1 31.2 76.1 28.2 post-none 306.3 99.1 18.0 5.0 20.4 7.1 31.1 12.2 65.1 32.1 88.9 41.8 86.2 39.8 post-fatigue 250 110.0 18.4 6.3 20.6 7.1 28.8 10.7 58.8 29.4 79.1 38.7 76.0 36.4 kinematic effect -8.4 -2.0 -3.2 0.28 1.54 2.78 0.96 p-value 0.51 0.21 0.14 0.2 0.84 0.75 0.91 fatigue effect -42.6 -0.2 -0.30 0.2 7.62 10.6 11.1 p-value 0.007 0.86 0.88 0.33 0.34 0.25 0.2 interaction effect -27.2 -0.4 0.36 0.08 -2.64 -1.6 -1.8 p-value 0.3 0.94 0.93 0.81 0.87 0.93 0.91 NCS group mean SD mean SD mean SD mean SD mean SD mean SD mean SD 156 pre-none 211.9 46.3 26.1 10.8 26.3 8.3 32.9 1.7 47.7 17.1 60.4 25.6 57.4 26.2

pre-fatigue 192.7 59.2 20.4 8.9 20.9 6.3 26.1 5.3 40.7 21.4 54.5 27.8 54.8 25.6 post-none 220 79.8 19.6 14.9 22.1 10.8 35.8 15.5 53.5 19.9 67.2 24.9 64.5 24.2 post-fatigue 207.3 78.1 16.8 14.1 18.6 10.9 32.9 17.2 50.9 22.2 65.9 27.7 63.9 25.9 kinematic effect 11.3 5.0 3.2 -0.68 -8.0 -9.06 -8.14 p-value 0.62 0.13 0.3 0.12 0.15 0.21 0.23 fatigue effect -15.9 4.2 4.4 0.2 4.8 3.7 1.62 p-value 0.49 0.197 0.16 0.62 0.37 0.6 0.8 interaction effect 6.56 -4.0 -1.6 -0.16 -4.44 -4.6 3.2 p-value 0.89 0.64 0.76 0.85 0.67 0.74 0.88

-0.48

Table C.3 . Linear mixed-model results for muscle force variables. Group means and standard deviations for each condition are listed along with the model effect and corresponding p-value. Analyses are performed separately for the group that experienced a meaningful reduction in core stability post-CSKP (CS group) and the group that did not (NCS group) in order to provide insight into which effects may be driven by a reduction in core stability. In the case where the best linear mixed model for a biomechanical parameter included only a kinematic effect, N/A is listed by the effects not included in the model. Significant effects (p<0.05) are bolded, trending effects m (0.5≤p≤0.1) are bolded and italicized. All muscle forces are normalized to the muscle group’s maximum isometric force property (F 0).

m m m m m m iGMax [%F 0] iGMed [%F 0] iGMin [%F 0] iIliacus [%F 0] iRF [%F 0] cMF [%F 0]

CS group mean SD mean SD mean SD mean SD mean SD mean SD pre-none 39.3 11.4 60.1 7.1 23.5 6.6 12.4 7.1 66.3 5.4 9.99 5.16 pre-fatigue 39.3 11.4 60.1 7.1 23.5 6.6 12.5 7.1 66.3 5.4 13.54 7.02 post-none 35.8 15.9 65.6 7.9 45.7 13.3 20.8 23.5 70.7 10.0 10.69 7.49 post-fatigue 35.9 15.9 65.6 7.9 45.7 13.3 20.8 23.5 70.7 10.0 14.69 11.9 kinematic effect -3.46 5.48 22.18 8.32 4.46 0.92 p-value 0.07 0.12 0.001 0.15 0.004 0.73 157 fatigue effect N/A N/A N/A N/A N/A 3.78

p-value N/A N/A N/A N/A N/A 0.18 interaction effect N/A N/A N/A N/A N/A 0.44 p-value N/A N/A N/A N/A N/A 0.93 NCS group mean SD mean SD mean SD mean SD mean SD mean SD pre-none 29.8 16.6 69.6 9.0 40.3 17.8 9.3 2.9 52.6 20.8 4.6 2.5 pre-fatigue 29.9 16.5 69.9 9.0 40.3 17.8 9.31 2.9 52.5 20.8 6.1 4.5 post-none 31.2 17.7 65.2 11.5 29.5 11.5 7.7 2.1 52.6 16.5 4.6 1.7 post-fatigue 31.3 17.7 65.2 11.5 29.5 11.5 7.6 2.1 52.4 16.5 6.2 6.7 kinematic effect 1.38 -4.42 -10.78 -1.66 -0.06 0.06 p-value 0.7 0.0002 0.026 0.02 0.99 0.97 fatigue effect N/A N/A N/A N/A N/A 1.58 p-value N/A N/A N/A N/A N/A 0.46 interaction effect N/A N/A N/A N/A N/A 0.12 p-value N/A N/A N/A N/A N/A 0.97 Abbreviations: Ipsilateral side (i), contralateral side (c), gluteus maximus (GMax), gluteus medius (GMed), gluteus minimus (GMin), rectus femoris (RF), multifidus (MF), quadratus lumborum (QL), psoas (PS), rectus abdominis (RA), internal oblique (IO), erector spinae (ES). Continued.

Table C.3 continued

m m m m cRA m iQL [%F 0] cQL [%F 0] cPS [%F 0] iRA [%F 0] m iIO [%F 0] [%F 0] CS group mean SD mean SD mean SD mean SD mean SD mean SD pre-none 5.7 2.84 5.7 2.84 1.3 1.6 32.2 4.7 39.2 8.6 32.9 10.2 pre-fatigue 8.4 5.82 8.4 5.82 1.2 1.3 19.7 5.8 23.6 6.3 28.4 8.0 post-none 16.2 20.5 16.2 20.5 0.7 0.5 26.5 9.5 28.3 9.6 30.8 9.0 post-fatigue 17.5 16.83 17.5 16.83 0.5 0.4 18.8 5.9 21.6 7.7 23.2 4.0 9.86 9.86 -0.64 -3.32 -6.42 -3.68 kinematic effect

p-value 0.12 0.12 0.06 0.34 0.16 0.13 fatigue effect 2.0 2.0 -0.1 -10.1 -11.1 -6.06 p-value 0.73 0.73 0.76 0.013 0.027 0.023 -1.4 -1.4 -0.12 4.76 8.92 -3.0 interaction effect

158 p-value 0.9 0.9 0.85 0.49 0.32 0.51 NCS group mean SD mean SD mean SD mean SD mean SD mean SD

pre-none 3.3 2.5 3.3 2.5 0.7 0.7 26.7 8.6 27 12.4 29.7 11.7 pre-fatigue 9.2 9.7 9.2 9.7 1.5 2.2 13.1 4.6 14.8 7.1 18.1 3.9 post-none 5.0 5.0 5.0 5.0 0.4 0.2 26.9 14.3 26.6 14.4 34.1 12.1 post-fatigue 7.1 7.1 7.1 7.1 0.9 1.0 11.2 2.6 10.6 2.2 17.6 2.3 kinematic effect -0.18 -0.18 -0.42 -0.90 -2.26 2.0 p-value 0.95 0.95 0.35 0.85 0.68 0.58 fatigue effect 4.0 4.0 0.62 -14.6 -14.1 -14.04 p-value 0.21 0.21 0.18 0.012 0.027 0.003 -3.84 -3.84 -0.28 -2.08 -3.8 -4.88 interaction effect p-value 0.52 0.52 0.76 0.83 0.73 0.5 Continued

Table C.3 continued

m m m m m cIO [%F 0] iEO [%F 0] cEO [%F 0] iES [%F 0] cES [%F 0]

CS group mean SD mean SD mean SD mean SD mean SD pre-none 31.7 5.7 13.8 4.3 14.0 3.6 30.2 22.6 41.9 22.1 pre-fatigue 27.2 3.7 10.8 3.8 14.4 3.5 27.4 10.1 36.5 10.8 post-none 28.3 2.8 15.3 2.8 14.8 3.6 34.4 18.9 38.5 26.3 post-fatigue 21.7 5.6 10.6 2.8 14.4 4.3 30.4 16.8 36.2 24.1 kinematic effect -4.46 0.68 0.34 3.2 -1.86

p-value 0.027 0.53 0.84 0.12 0.49 fatigue effect -5.58 -3.84 -0.004 -3.4 -3.82 p-value 0.009 0.005 0.99 0.14 0.17 interaction effect -2.04 -1.6 -0.76 -1.32 3.04

159 p-value 0.56 0.46 0.82 0.76 0.57 NCS group mean SD mean SD mean SD mean SD mean SD

pre-none 25.3 9.9 10.7 2.9 18.1 4.7 20.4 6.1 22.7 15.8 pre-fatigue 16 7.3 8.3 3.4 15.8 5.1 15.8 7.6 20.5 7.3 post-none 25.9 12.3 11.6 4.7 14.9 3.3 17.6 3.6 23.1 13.3 post-fatigue 14.2 6.6 7.2 1.2 12.9 4.0 15.8 3.7 21.0 11.6 kinematic effect -0.6 -0.1 -3.02 -1.42 0.42

p-value 0.84 0.94 0.18 0.61 0.93 fatigue effect -10.44 -3.4 -2.2 -3.2 -2.2 p-value 0.007 0.033 0.31 0.26 0.67 interaction effect -2.36 -2.12 0.20 2.88 0.24

p-value 0.7 0.45 0.958 0.6 0.98

Appendix D : Individual Participant Data for Subject-Specific Simulations

160

Table D.1. Individual participant data for subject-specific simulations. Individual changes in core stability (CoPexc), core muscle fatigue (MedF), and main lower extremity kinematic changes from pre-CSKP to post-CSKP jogging. Kinematic changes were determined from experimental data using custom Vicon, Body Builder, and MATLAB codes prior to bringing the data into OpenSim and creating jogging simulations. All angles represent the peak angle over the dominant-leg stance phase. Mass Height Age %change %change Main kinematic changes: average pre- Sex (kg) (m) (y) CoPexc MedF CSKP, average post-CSKP CS group hAddA:, 10.8°, 14.4° P46 62.65 1.73 18 M 18 -28 kAbdA: -2.9°, -5.2° hIRA: 8.0°, 10.9° P51 76.37 1.88 26 M 11 -19 kAbdA: -4.6°,-2.9° hAddA: 4.9°, 3.5°; hIRA: 12.2°, 9.7° P88 70.55 1.76 21 M 26 -20 kAbdA: 0.05°, -2.1°; kIRA: 11.3°, 15.5° 161 hIRA: -1.7°, 1.9°

P42 90.60 1.88 21 M 38 -6 kAbdA: -3.4°, -2.5° NCS group hAddA: 9.8°, 11.8° P54 100.97 1.85 30 M -12 -25 kAbdA: 0.8°, 0.4° hIRA: 8.1°, 9.8° P82 54.40 1.63 22 F 9 -19 kIRA: 9.7°, 8.0° P63 57.92 1.71 20 F 9 -28 kAbdA: -2.1°, -3.3°; kIRA: 10.5°, 14.6° hAddA: 16.6°, 12.7°; hIRA: 16.6°, 14.9° P16 59.27 1.70 52 F 5 -34 kAbdA: -4.6°, -5.3°; kIRA: 12.1°, 13.3° Abbreviations: hip adduction angle (hAddA), hip internal rotation angle (hIRA), knee abduction angle (kAbdA), knee internal rotation angle (kIRA), center of pressure excursion during quiet sitting (CoPexc), average core muscle median frequency (MedF), core stability knockdown protocol (CSKP), group with reduced core stability (CoPexc>10%) post-CSKP (CS group), group with no change in core stability post-CSKP (NCS group

Appendix E : Residual and Reserve Results for OpenSim Simulations

Figure E.1. Simulation quality thresholds established for residual forces and moments and reserve actuators in OpenSim.

162

Figure E.2. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 1.

163

Figure E.3. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 2.

164

Figure E.4. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 3.

165

Figure E.5. Residual forces and moments and reserve actuators added to the model for participant P46 in simulation 4.

166

Figure E.6. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 1.

167

Figure E.7. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 2.

168

Figure E.8. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 3.

169

Figure E.9. Residual forces and moments and reserve actuators added to the model for participant P51 in simulation 4.

170

Figure E.10. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 1.

171

Figure E.11. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 2.

172

Figure E.12. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 3.

173

Figure E.13. Residual forces and moments and reserve actuators added to the model for participant P88 in simulation 4.

174

Figure E.14. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 1.

175

Figure E.15. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 2.

176

Figure E.16. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 3.

177

Figure E.17. Residual forces and moments and reserve actuators added to the model for participant P42 in simulation 4.

178

Figure E.18. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 1.

179

Figure E.19. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 2.

180

Figure E.20. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 3.

181

Figure E.21. Residual forces and moments and reserve actuators added to the model for participant P54 in simulation 4.

182

Figure E.22. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 1.

183

Figure E.23. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 2.

184

Figure E.24. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 3.

185

Figure E.25. Residual forces and moments and reserve actuators added to the model for participant P63 in simulation 4.

186

Figure E.26. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 1.

187

Figure E.27. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 2.

188

Figure E.28. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 3.

189

Figure E.29. Residual forces and moments and reserve actuators added to the model for participant P82 in simulation 4.

190

Figure E.30. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 1.

191

Figure E.31. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 2.

192

Figure E.32. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 3.

193

Figure E.33. Residual forces and moments and reserve actuators added to the model for participant P16 in simulation 4.

194

References

Alexander, R. M. (1984). "Walking and running: Legs and leg movements are subtly

adapted to minimize the energy costs of locomotion." American Scientist 72(4):

348-354.

Allison, G. and T. Fujiwara (2002). "The relationship between EMG median frequency

and low frequency band amplitude changes at different levels of muscle

capacity." Clinical Biomechanics 17(6): 464-469.

Andriacchi, T., E. Alexander, et al. (1998). "A point cluster method for in vivo analysis:

applied to a study of knee kinematics." J. Biomech. Eng. 120(6): 743-749.

Andriacchi, T., G. Kramer, et al. (1985). The biomechanics of running and knee injuries.

Academy of Orthopaedic Surgeons Symposium on Sports Medicine: The Knee.,

St. Louis.

Andriacchi, T. P., E. J. Alexander, et al. (1998). "A point cluster method for in vivo

motion analysis: applied to a study of knee kinematics." Journal of Biomechanical

Engineering 120(6): 743-749.

Andriacchi, T. P., P. L. Briant, et al. (2006). "Rotational changes at the knee after ACL

injury cause cartilage thinning." Clinical Orthopaedics and Related Research 442:

39-44.

195

Andriacchi, T. P. and C. O. Dyrby (2005). "Interactions between kinematics and loading

during walking for the normal and ACL deficient knee." Journal of Biomechanics

38(2): 293-298.

Arnold, E. M., S. R. Hamner, et al. (2013). "How muscle fiber lengths and velocities

affect muscle force generation as humans walk and run at different speeds." J Exp

Biol 216(Pt 11): 2150-2160.

Arnold, E. M., S. R. Ward, et al. (2010). "A model of the lower limb for analysis of

human movement." Annals of biomedical engineering 38(2): 269-279.

Barton, C. J., S. Lack, et al. (2013). "Gluteal muscle activity and patellofemoral pain

syndrome: a systematic review." Br J Sports Med 47(4): 207-214.

Besier, T. F., M. Fredericson, et al. (2009). "Knee muscle forces during walking and

running in patellofemoral pain patients and pain-free controls." J Biomech 42(7):

898-905.

Bilodeau, M., S. Schindler-Ivens, et al. (2003). "EMG frequency content changes with

increasing force and during fatigue in the quadriceps femoris muscle of men and

women." J Electromyogr Kinesiol 13(1): 83-92.

Bouisset, S. (1991). "Relationship between postural support and intentional movement:

biomechanical approach." Arch Int Physiol Biochim Biophys 99(5): A77-92.

Brown, T., M. O’Donovan, et al. (2014). "Body borne loads impact walk-to-run and

running biomechanics." Gait & Posture 40(1): 237-242.

196

Bruno, A. G., D. E. Anderson, et al. (2012). "The effect of thoracic kyphosis and sagittal

plane alignment on vertebral compressive loading." J Bone Miner Res 27(10):

2144-2151.

Bruno, A. G., M. L. Bouxsein, et al. (2015). "Development and Validation of a

Musculoskeletal Model of the Fully Articulated Thoracolumbar Spine and Rib

Cage." J Biomech Eng 137(8): 081003.

Buff, H. U., L. C. Jones, et al. (1988). "Experimental determination of forces transmitted

through the patello-femoral joint." Journal of Biomechanics 21(1): 17-23.

Buford, W. L., F. M. Ivey, et al. Moment Arms of Muscles at the Knee. Orthopaedic

Research Society, New Orleans, Louisiana.

Buist, I., S. W. Bredeweg, et al. (2010). "Incidence and risk factors of running-related

injuries during preparation for a 4-mile recreational running event." Br J Sports

Med 44(8): 598-604.

Buist, I., S. W. Bredeweg, et al. (2008). "No effect of a graded training program on the

number of running-related injuries in novice runners: a randomized controlled

trial." Am J Sports Med 36(1): 33-39.

Bullock-Saxton, J. E., V. Janda, et al. (1994). "The influence of ankle sprain injury on

muscle activation during hip extension." International Journal of Sports Medicine

15(6): 330-334.

Burnham, K. P. and D. R. Anderson (2004). "Multimodel inference understanding AIC

and BIC in model selection." Sociological methods & research 33(2): 261-304.

197

Calisse, J., A. Rohlmann, et al. (1999). "Estimation of trunk muscle forces using the finite

element method and in vivo loads measured by telemeterized internal spinal

fixation devices." Journal of Biomechanics 32(7): 727-731.

Callaghan, J. P. and S. M. McGill (2001). "Low back joint loading and kinematics during

standing and unsupported sitting." Ergonomics 44(3): 280-294.

Cappellini, G., Y. P. Ivanenko, et al. (2006). "Motor patterns in human walking and

running." J Neurophysiol 95(6): 3426-3437.

Caruthers, E., J. A. Thompson, et al. (2013). Individual Muscle Forces During a Sit to

Stand Transfer. Annual Meeting of the American Society of Biomechanics,

Omaha, NE.

Cavanagh, P. R. and K. R. Williams (1982). "The effect of stride length variation on

oxygen uptake during distance running." Medicine and Science in Sports and

Exercise 14(1): 30-35.

Chaudhari, A. W., S. Jamison, et al. (2012). Proximal Risk Factors for ACL Injury: Role

of Core Stability. ACL Injuries in the Female Athlete. F. R. Noyes and S. Barber-

Westin, Springer Berlin Heidelberg: 169-183.

Chen, Y. J. and C. M. Powers (2014). "Comparison of three-dimensional patellofemoral

joint reaction forces in persons with and without patellofemoral pain." J Appl

Biomech 30(4): 493-500.

Cholewicki, J., K. Juluru, et al. (1999). "Intra-abdominal pressure mechanism for

stabilizing the lumbar spine." J Biomech 32(1): 13-17.

198

Cholewicki, J. and S. M. McGill (1996). "Mechanical stability of the in vivo lumbar

spine: implications for injury and chronic low back pain." Clin Biomech (Bristol,

Avon) 11(1): 1-15.

Cholewicki, J., G. K. Polzhofer, et al. (2000). "Postural control of trunk during unstable

sitting." J Biomech 33(12): 1733-1737.

Christophy, M. (2010). A detailed open-source musculoskeletal model of the human

lumbar spine, University of California, Berkeley.

Christophy, M., N. Faruk Senan, et al. (2012). "A musculoskeletal model for the lumbar

spine." Biomechanics and modeling in mechanobiology 11(1-2): 19-34.

Cichanowski, H. R., J. S. Schmitt, et al. (2007). "Hip strength in collegiate female

athletes with patellofemoral pain." Medicine and Science in Sports and Exercise

39(8): 1227-1232.

Cowan, S. M., K. M. Crossley, et al. (2009). "Altered hip and trunk muscle function in

individuals with patellofemoral pain." Br J Sports Med 43(8): 584-588.

Creps, J. (2014). An Investigation of Simulated Core Muscle Activation during Running

and its Effect on Knee Loading and Lower Extremity Muscle Activation Using

OpenSim. Master of Science, The Ohio State Univeristy.

Crisco, J. J., M. M. Panjabi, et al. (1992). "Euler stability of the human ligamentous

lumbar spine. Part II: Experiment." Clin Biomech (Bristol, Avon) 7(1): 27-32.

Crowell, H. P. and I. S. Davis (2011). "Gait retraining to reduce lower extremity loading

in runners." Clinical Biomechanics 26(1): 78-83.

199

Crowninshield, R. D. and R. A. Brand (1981). "A physiologically based criterion of

muscle force prediction in locomotion." J Biomech 14(11): 793-801.

Daoud, A. I., G. J. Geissler, et al. (2012). "Foot strike and injury rates in endurance

runners: a retrospective study." Med Sci Sports Exerc 44(7): 1325-1334.

Davis, I. S., B. J. Bowser, et al. (2015). "Greater vertical impact loading in female

runners with medically diagnosed injuries: a prospective investigation." British

Journal of Sports Medicine.

De Luca, C. J. (1983). "Myoelectrical manifestations of localized muscular fatigue in

humans." Critical reviews in biomedical engineering 11(4): 251-279. de Zee, M., L. Hansen, et al. (2007). "A generic detailed rigid-body lumbar spine model."

Journal of Biomechanics 40(6): 1219-1227.

Delp, S., F. Anderson, et al. (2007). "OpemSim: open-source software to create and

analyze dynamic simulations of movement." IEEE Trans. Biomed. Eng. 54(11):

1940-1950.

Delp, S., S. Suryanarayanan, et al. (2001). "Architecture of the rectus abdominis,

quadratus lumborum, and erector spinae." Journal of Biomechanics 34(3): 371-

375.

Dempsey, A. R., D. G. Lloyd, et al. (2007). "The effect of technique change on knee

loads during sidestep cutting." Medicine and Science in Sports and Exercise

39(10): 1765-1773.

Dempster, W. T. and G. R. L. Gaughran (1967). "Properties of body segments based on

size and weight." Am. J. Anat. American Journal of Anatomy 120(1): 33-54.

200

Dierks, T. A., K. T. Manal, et al. (2008). "Proximal and distal influences on hip and knee

kinematics in runners with patellofemoral pain during a prolonged run." Journal

of Orthopaedic and Sports Physical Therapy 38(8): 448-456.

Donnelly, C., D. G. Lloyd, et al. (2012). "Optimizing whole-body kinematics to minimize

valgus knee loading during sidestepping: implications for ACL injury risk."

Journal of Biomechanics 45(8): 1491-1497.

Downey, R., R. Kamalapurkar, et al. (2016). Compensating for Fatigue-Induced Time-

Varying Delayed Muscle Response in Neuromuscular Electrical Stimulation

Control. Recent Results on Nonlinear Delay Control Systems: In honor of

Miroslav Krstic. I. Karafyllis, M. Malisoff, F. Mazenc and P. Pepe. Cham,

Springer International Publishing: 143-161.

Drechsler, W. I., M. C. Cramp, et al. (2006). "Changes in muscle strength and EMG

median frequency after anterior cruciate ligament reconstruction." Eur J Appl

Physiol 98(6): 613-623.

El-Rich, M., A. Shirazi-Adl, et al. (2004). "Muscle activity, internal loads, and stability

of the human spine in standing postures: combined model and in vivo studies."

Spine 29(23): 2633-2642.

Fairclough, J., K. Hayashi, et al. (2006). "The functional anatomy of the iliotibial band

during flexion and extension of the knee: implications for understanding iliotibial

band syndrome." Journal of Anatomy 208(3): 309-316.

201

Farrokhi, S., P. M. Colletti, et al. (2011). "Differences in patellar cartilage thickness,

transverse relaxation time, and deformational behavior: a comparison of young

women with and without patellofemoral pain." Am J Sports Med 39(2): 384-391.

Ferber, R., L. Bolgla, et al. (2015). "Strengthening of the hip and core versus knee

muscles for the treatment of patellofemoral pain: a multicenter randomized

controlled trial." J Athl Train 50(4): 366-377.

Ferber, R., A. Hreljac, et al. (2009). "Suspected Mechanisms in the Cause of Overuse

Running Injuries: A Clinical Review." Sports Health 1(3): 242-246.

Ferber, R., B. Noehren, et al. (2010). "Competitive female runners with a history of

iliotibial band syndrome demonstrate atypical hip and knee kinematics." J Orthop

Sports Phys Ther 40(2): 52-58.

Fisher, S. T. (2010). The intra-session and inter-session reliability of centre-of-pressure

based measures of postural sway within a normal population Master of

Osteopathy Thesis, UNITEC Institute of Technology.

Fox, A. J., F. Wanivenhaus, et al. (2012). "The basic science of the patella: structure,

composition, and function." J Knee Surg 25(2): 127-141.

Fredericson, M., C. L. Cookingham, et al. (2000). "Hip abductor weakness in distance

runners with iliotibial band syndrome." Clinical Journal of Sport Medicine 10(3):

169-175.

Fredericson, M. and T. Moore (2005). "Muscular balance, core stability, and injury

prevention for middle- and long-distance runners." Physical Medicine and

Rehabilitation Clinics of North America 16 (3): 669-689.

202

Fredericson, M. and A. Tenforde (2016). Running Injuries, An Issue of Physical

Medicine and Rehabilitation Clinics of North America, Elsevier Health Sciences.

Freeman, M. D., M. A. Woodham, et al. (2010). "The role of the lumbar multifidus in

chronic low back pain: a review." PM R 2(2): 142-146; quiz 141 p following 167.

Gates, D. H. and J. B. Dingwell (2010). "Muscle fatigue does not lead to increased

instability of upper extremity repetitive movements." Journal of Biomechanics

43(5): 913-919.

Gates, D. H. and J. B. Dingwell (2011). "The effects of muscle fatigue and movement

height on movement stability and variability." Exp Brain Res 209(4): 525-536.

Granata, K. P. and P. Gottipati (2008). "Fatigue influences the dynamic stability of the

torso." Ergonomics 51(8): 1258-1271.

Granata, K. P., G. P. Slota, et al. (2004). "Influence of Fatigue in Neuromuscular Control

of Spinal Stability." Human Factors 46(1): 81-91.

Hamner, S., A. Seth, et al. (2010). "Muscle contributions to propulsion and support

during running." J Biomech 43(14): 2709-2716.

Harrast, M. A. and D. Colonno (2010). "Stress Fractures in Runners." Clinics in Sports

Medicine 29(3): 399-416.

Hart, J. M., J. M. Fritz, et al. (2006). "Reduced Quadriceps Activation After Lumbar

Paraspinal Fatiguing Exercise." Journal of Athletic Training 41(1): 79-86.

Hart, J. M., D. C. Kerrigan, et al. (2009). "Jogging kinematics after lumbar paraspinal

muscle fatigue." J Athl Train 44(5): 475-481.

203

Heino Brechter, J. and C. M. Powers (2002). "Patellofemoral stress during walking in

persons with and without patellofemoral pain." Med Sci Sports Exerc 34(10):

1582-1593.

Hendershot, B. D. and M. A. Nussbaum (2013). "Persons with lower-limb amputation

have impaired trunk postural control while maintaining seated balance." Gait

Posture 38(3): 438-442.

Hicks, J. (2011). "OpenSim User's Guide." 2013, from http://simtk-

confluence.stanford.edu:8080/display/OpenSim/User%27s+Guide.

Hicks, J. and C. Dembia. (2014). "How Static Optimization Works." from http://simtk-

confluence.stanford.edu:8080/display/OpenSim/How+Static+Optimization+Work

Hicks, J. L., T. K. Uchida, et al. (2015). "Is my model good enough? Best practices for

verification and validation of musculoskeletal models and simulations of

movement." J Biomech Eng 137(2).

Hides, J. A., W. R. Stanton, et al. (2008). "Effect of stabilization training on multifidus

muscle cross-sectional area among young elite cricketers with low back pain."

Journal of Orthopaedic and Sports Physical Therapy 38(3): 101-108.

Hodges, P., A. K. Holm, et al. (2006). "Rapid atrophy of the lumbar multifidus follows

experimental disc or nerve root injury." Spine (Phila Pa 1976) 31(25): 2926-2933.

Hodges, P. W. (1999). "Is there a role for transversus abdominis in lumbo-pelvic

stability?" Manual therapy 4(2): 74-86.

204

Hodges, P. W. and C. A. Richardson (1997). "Contraction of the Abdominal Muscles

Associated with Movement of the Lower Limb." Physical Therapy 77(2): 132-

144.

Hodges, P. W. and C. A. Richardson (1997). "Relationship between limb movement

speed and associated contraction of the trunk muscles." Ergonomics 40(11): 1220-

1230.

Holzbaur, K. R., W. M. Murray, et al. (2005). "A model of the upper extremity for

simulating musculoskeletal surgery and analyzing neuromuscular control." Ann

Biomed Eng 33 (6): 829-840.

Huberti, H. H. and W. C. Hayes (1984). "Patellofemoral contact pressures. The influence

of q-angle and tendofemoral contact." J Bone Joint Surg Am 66(5): 715-724.

Ireland, M., J. Willson, et al. (2003). "Hip strength in females with and without

patellofemoral pain." Journal of Orthopaedic and Sports Physical Therapy 33:

671-676.

James, S. L. (1995). "Running injuries to the knee." Journal of the American Academy of

Orthopaedic Surgeons 3(6): 309-318.

Jamison, S. and A. Chaudhari (2009). "The Influence of Pelvic Control on Running

Mechanics." American Society of Biomechanics Meeting.

Jamison, S. T., M. P. McNally, et al. (2013). "The effects of core muscle activation on

dynamic trunk position and knee abduction moments: implications for ACL

injury." J Biomech 46(13): 2236-2241.

205

Jamison, S. T., R. J. McNeilan, et al. (2012). "Randomized controlled trial of the effects

of a trunk stabilization program on trunk control and knee loading." Med Sci

Sports Exerc 44 (10): 1924-1934.

Jamison, S. T., X. Pan, et al. (2012). "Knee moments during run-to-cut maneuvers are

associated with lateral trunk positioning." J Biomech 45(11): 1881-1885.

Jorgensen, M. J., W. S. Marras, et al. (2001). "MRI-derived moment-arms of the female

and male spine loading muscles." Clin Biomech (Bristol, Avon) 16(3): 182-193.

Kearney, J. T. and G. A. Stull (1981). "Effect of fatigue level on rate of force

development by the grip-flexor muscles." Medicine and Science in Sports and

Exercise 13(5): 339-342.

Keller, T. S. and A. L. Roy (2002). "Posture-dependent isometric trunk extension and

flexion strength in normal male and female subjects." J Spinal Disord Tech 15(4):

312-318.

Khalaf, K. A., M. Parnianpour, et al. (1997). "Modeling of functional trunk muscle

performance: interfacing ergonomics and spine rehabilitation in response to the

ADA." J Rehabil Res Dev 34(4): 459-469.

Kibler, W., J. Press, et al. (2006). "The role of core stability in athletic function." Sports

Med 36(3): 189-198.

Kumar, S., R. M. Dufresne, et al. (1995). "Human trunk strength profile in lateral flexion

and axial rotation." Spine (Phila Pa 1976) 20(2): 169-177.

206

Lafond, D., H. Corriveau, et al. (2004). "Intrasession reliability of center of pressure

measures of postural steadiness in healthy elderly people." Arch Phys Med

Rehabil 85(6): 896-901.

Lavine, R. (2010). "Iliotibial band friction syndrome." Current Reviews in

Musculoskeletal Medicine 3(1-4): 18-22.

Lee, T., S. Anzel, et al. (1994). "The influence of fixed rotational deformities of the

femur on the patellorfemoral contact pressures in human cadaver knees." Clin.

Othop. Relat. Res. 302: 69-74.

Lenhart, R. L., D. G. Thelen, et al. (2014). "Increasing running step rate reduces

patellofemoral joint forces." Med Sci Sports Exerc 46(3): 557-564.

Lenth, R. V. (2006). Java Applets for Power and Sample Size.

Lin, Y. C., T. W. Dorn, et al. (2012). "Comparison of different methods for estimating

muscle forces in human movement." Proc Inst Mech Eng H 226(2): 103-112.

Lund, K. and C. Dembia. (2013). "Getting Started With CMC." Retrieved September

24, 2014, from http://simtk-

confluence.stanford.edu:8080/display/OpenSim/Getting+Started+with+CMC.

Macera, C., K. Powell, et al. (1989). "Predicting lower-extremity injuries among habitual

runners." Archives of internal medicine 149(11): 2565-2568.

MacMahon, J., A. Chaudhari, et al. (2000). Biomechanical injury predictors for marathon

runners: striding toward iliotibial band syndrome injury prevention. XXVIII

Intenational Symposium of Biomechanics in Sports. Hong Kong.

207

Madigan, M. L., B. S. Davidson, et al. (2006). "Postural sway and joint kinematics during

quiet standing are affected by lumbar extensor fatigue." Hum Mov Sci 25(6): 788-

799.

Maki, B. E., P. J. Holliday, et al. (1990). "Aging and Postural Control : A Comparison of

Spontaneous- and Induced-Sway Balance Tests." Journal of the American

Geriatrics Society 38(1): 1-9.

Marti, B., J. P. Vader, et al. (1988). "On the epidemiology of running injuries. The 1984

Bern Grand-Prix study." Am J Sports Med 16(3): 285-294.

Matheson, G. O., D. B. Clement, et al. (1987). "Stress fractures in athletes. A study of

320 cases." Am J Sports Med 15(1): 46-58.

McGill, S. (2010). "Core training: Evidence translating to better performance and injury

prevention." Strength & Conditioning Journal 32(3): 33-46.

McGill, S., D. Juker, et al. (1996). "Appropriately placed surface EMG electrodes reflect

deep muscle activity (psoas, quadratus lumborum, abdominal wall) in the lumbar

spine." Journal of Biomechanics 29(11): 1503-1507.

Méjane, J., J. Faubert, et al. (2016). EVALUATING THE EFFECT OF A

PERCEPTUAL-COGNITIVE TASK ON LANDING BIOMECHANICS OF

THE LOWER LIMB. ISBS-Conference Proceedings Archive.

Messier, S. P., S. E. Davis, et al. (1991). "Etiologic factors associated with patellofemoral

pain in runners." Med Sci Sports Exerc 23(9): 1008-1015.

208

Milner, C. E., R. Ferber, et al. (2006). "Biomechanical factors associated with tibial stress

fracture in female runners." Medicine and science in sports and exercise 38(2):

323-328.

Mizuno, Y., M. Kumagai, et al. (2001). "Q-angle influences tibiofemoral and

patellofemoral kinematics." J Orthop Res 19(5): 834-840.

Moghadam, M., H. Ashayeri, et al. (2011). "Reliability of center of pressure measures of

postural stability in healthy older adults: Effects of postural task difficulty and

cognitive load." Gait & Posture 33(4): 651-655.

Mokhtarzadeh, H., L. Perraton, et al. (2014). "A comparison of optimisation methods and

knee joint degrees of freedom on muscle force predictions during single-leg hop

landings." J Biomech 47(12): 2863-2868.

Monaghan, K., E. Delahunt, et al. (2006). "Ankle function during gait in patients with

chronic ankle instability compared to controls." Clinical Biomechanics 21(2):

168-174.

Nachemson, A. and J. M. Morris (1964). "In vivo measurements of intradiscal pressure."

The Journal of Bone & Joint Surgery 46(5): 1077-1092.

Neumann, D. A. (2013). Kinesiology of the musculoskeletal system: foundations for

rehabilitation, Elsevier Health Sciences.

Noble, C. A. (1980). "Iliotibial band friction syndrome in runners." Am J Sports Med

8(4): 232-234.

209

Noehren, B., I. Davis, et al. (2007). "ASB Clinical Biomechanics Award Winner 2006:

Prospective study of the biomechanical factors assiciated with iliotibial band

syndrome." Clinical Biomechanics 22: 951-956.

Noehren, B., J. Hamill, et al. (2013). "Propspective Evidence for a Hip Etiology in

Patellofemoral Pain." Med. Sci. Sports Exerc. 45(6): 1120-1124.

Noehren, B., J. Scholz, et al. (2010). "The effect of real-time gait retraining on hip

kinematics, pain and function in subjects with patellofemoral pain syndrome."

British Journal of Sports Medicine: bjsports69112.

Novacheck, T. F. (1998). "The biomechanics of running." Gait & Posture 7(1): 77-95.

Orchard, J. W., P. A. Fricker, et al. (1996). "Biomechanics of iliotibial band friction

syndrome in runners." Am J Sports Med 24(3): 375-379.

Paasuke, M., J. Ereline, et al. (1999). "Neuromuscular fatigue during repeated exhaustive

submaximal static contractions of knee extensor muscles in endurance-trained,

power-trained and untrained men." Acta Physiologica Scandinavica 166(4): 319-

326.

Panjabi, M. M. (1992). "The stabilizing system of the spine. Part I. Function, dysfunction,

adaptation, and enhancement." J Spinal Disord 5(4): 383-389; discussion 397.

Parnianpour, M., M. Nordin, et al. (1988). "1988 Volvo award in biomechanics. The

triaxial coupling of torque generation of trunk muscles during isometric exertions

and the effect of fatiguing isoinertial movements on the motor output and

movement patterns." Spine (Phila Pa 1976) 13(9): 982-992.

210

Pohl, M. B., D. R. Mullineaux, et al. (2008). "Biomechanical predictors of retrospective

tibial stress fractures in runners." J Biomech 41(6): 1160-1165.

Potvin, J. and P. R. O'Brien (1998). "Trunk muscle co-contraction increases during

fatiguing, isometric, lateral bend exertions. Possible implications for spine

stability." Spine 23(7): 774-780.

Powers, C. M. (2003). "The influence of altered lower-extremity kinematics on

patellofemoral joint dysfunction: a theoretical perspective." Journal of

Orthopaedic and Sports Physical Therapy 33(11): 639-646.

Powers, C. M. (2010). "The influence of abnormal hip mechanics on knee injury: a

biomechanical perspective." Journal of Orthopaedic and Sports Physical Therapy

40(2): 42-51.

Raabe, M. E. and A. M. W. Chaudhari (2016). "An Investigation of Jogging

Biomechanics using the Full-Body Lumbar Spine Model: Model Development

and Validation." Journal of Biomechanics 49(7): 1238-1243.

Raabe, M. E., S. M. Monfort, et al. (2015). "Development of a Novel Protocol to

Temporarily Reduce Core Stability." Proceedings of 2015 ASB 39(452).

Raabe, M. E., S. M. Monfort, et al. (2016). "Isolated Core Muscle Fatigue Does Not

Generally Affect Core Stability." Proceedings of 2016 ASB 40(400).

Raabe, M. E., R. Teater, et al. (2016). "Intra- and Inter-Rater Reliability of a Novel Core

Stability Test." Proceedings of 2016 ASB 40(209).

211

Radebold, A., J. Cholewicki, et al. (2001). "Impaired postural control of the lumbar spine

is associated with delayed muscle response times in patients with chronic

idiopathic low back pain." Spine (Phila Pa 1976) 26(7): 724-730.

Rixe, J. A., R. A. Gallo, et al. (2012). "The barefoot debate: can minimalist shoes reduce

running-related injuries?" Curr Sports Med Rep 11(3): 160-165.

Rohlmann, A., L. Bauer, et al. (2006). "Determination of trunk muscle forces for flexion

and extension by using a validated finite element model of the lumbar spine and

measured in vivo data." Journal of Biomechanics 39(6): 981-989.

Romani, W. A., J. H. Gieck, et al. (2002). "Mechanisms and Management of Stress

Fractures in Physically Active Persons." Journal of Athletic Training 37(3): 306-

314.

Roy, S. and C. De Luca (1989). "Evolving characteristics of the median frequency of the

EMG signal." Computer-aided electromyography and expert systems. Holland:

Elsevier: 115-124.

Rudolph, K. S., M. J. Axe, et al. (2001). "Dynamic stability in the anterior cruciate

ligament deficient knee." Knee surgery, sports traumatology, arthroscopy 9(2):

62-71.

Running, U. S. A. (2015, July 13, 2015). "2015 State of the Sport - U.S. Race Trends."

Retrieved April 13, 2016, from http://www.runningusa.org/2015-state-of-sport-

us-trends.

Sallis, J. F., M. F. Hovell, et al. (1992). "Predictors of adoption and maintenance of

vigorous physical activity in men and women." Prev Med 21(2): 237-251.

212

Salsich, G. B., J. H. Brechter, et al. (2001). "Lower extremity kinetics during stair

ambulation in patients with and without patellofemoral pain." Clin Biomech

(Bristol, Avon) 16(10): 906-912.

Schmitz, A., K. Russo, et al. (2014). "Do novice runners have weak hips and bad running

form?" Gait Posture 40(1): 82-86.

Schultz, A., K. Haderspeck, et al. (1983). "Use of lumbar trunk muscles in isometric

performance of mechanically complex standing tasks." Journal of Orthopaedic

Research 1(1): 77-91.

Scott, S. H. and D. A. Winter (1990). "Internal forces of chronic running injury sites."

Medicine and science in sports and exercise 22(3): 357-369.

Shirazi, Z., M. Moghaddam, et al. (2014). "Comparative evaluation of core muscle

recruitment pattern in response to sudden external perturbations in patients with

patellofemoral pain syndrome and healthy subjects." Arch Phys Med Rehabil

95(7): 1383-1389.

Singh, N. B., A. Arampatzis, et al. (2010). "Effect of fatigue on force fluctuations in knee

extensors in young adults." Philos Trans A Math Phys Eng Sci 368(1920): 2783-

2798.

Sparrow, W. A. (2000). Energetics of human activity, Human Kinetics.

Sparto, P. J., M. Parnianpour, et al. (1997). "Neuromuscular trunk performance and

spinal loading during a fatiguing isometric trunk extension with varying torque

requirements." Journal of Spinal Disorders 10 (2): 145-156.

213

Steele, K. M., M. S. Demers, et al. (2012). "Compressive tibiofemoral force during

crouch gait." Gait Posture 35(4): 556-560.

Stefanyshyn, D. J., P. Stergiou, et al. (2006). "Knee angular impulse as a predictor of

patellofemoral pain in runners." Am J Sports Med 34(11): 1844-1851.

Sung, P. S. (2003). "Multifidi muscles median frequency before and after spinal

stabilization exercises." Archives of Physical Medicine and Rehabilitation 84(9):

1313-1318.

Szucs, K., A. Navalgund, et al. (2009). "Scapular muscle activation and co-activation

following a fatigue task." Med Biol Eng Comput 47(5): 487-495.

Tateuchi, H., S. Shiratori, et al. (2015). "The effect of angle and moment of the hip and

knee joint on iliotibial band hardness." Gait and Posture 41(2): 522-528.

Taunton, J., M. Ryan, et al. (2002). "A retrospective case-control analysis of 2002

running injuries." British Journal of Sports Medicine 36(2): 95-101.

Thelen, D. G. and F. C. Anderson (2006). "Using computed muscle control to generate

forward dynamic simulations of human walking from experimental data." J

Biomech 39(6): 1107-1115.

Thompson, J. A., A. M. Chaudhari, et al. (2013). "Gluteus maximus and soleus

compensate for simulated quadriceps atrophy and activation failure during

walking." J Biomech 46(13): 2165-2172.

Tonoli, C., E. Cumps, et al. (2010). "Incidence, risk factors and prevention of running

related injuries in long-distance running: A systematic review injury, location and

type." Sport Geneeskd. Sport en Geneeskunde 43(5): 12-18.

214

Umberger, B. R. (2010). "Stance and swing phase costs in human walking." J R Soc

Interface 7(50): 1329-1340.

Umberger, B. R., K. G. Gerritsen, et al. (2003). "A model of human muscle energy

expenditure." Comput Methods Biomech Biomed Engin 6(2): 99-111. van der Burg, J. C., E. E. van Wegen, et al. (2006). "Postural control of the trunk during

unstable sitting in Parkinson's disease." Parkinsonism Relat Disord 12(8): 492-

498. van der Krogt, M., S. Delp, et al. (2012). "How robust is human gait to muscle

weakness?" Gait & Posture 36(1): 113-119. van Dieen, J. H., T. Luger, et al. (2012). "Effects of fatigue on trunk stability in elite

gymnasts." Eur J Appl Physiol 112(4): 1307-1313. van Mechelen, W. (1992). "Running injuries. A review of the epidemiological literature."

Sports Med 14(5): 320-335.

Verhagen, E. (2012). "Prevention of running-related injuries in novice runners: are we

running on empty?" British journal of sports medicine 46(12): 836-837.

Wallwork, T. L., W. R. Stanton, et al. (2009). "The effect of chronic low back pain on

size and contraction of the lumbar multifidus muscle." Man Ther 14(5): 496-500.

Wilke, H.-J., A. Rohlmann, et al. (2003). "ISSLS prize winner: a novel approach to

determine trunk muscle forces during flexion and extension: a comparison of data

from an in vitro experiment and in vivo measurements." Spine 28(23): 2585-2593.

215

Willson, J., C. Dougherty, et al. (2005). "Core stability and its relationship to lower

extremity function and injury." The Journal of the American Academy of

Orthopaedic Surgeons 13(5): 316-325.

Willson, J., T. Kernozek, et al. (2011). "Gluteal muscle activation during running in

females with and without patellofemoral pain syndrome." Clinical Biomechanics

26(7): 735-740.

Willson, J. D., S. Binder-Macleod, et al. (2008). "Lower extremity jumping mechanics of

female athletes with and without patellofemoral pain before and after exertion."

Am J Sports Med 36(8): 1587-1596.

Willson, J. D. and I. S. Davis (2008). "Lower extremity mechanics of females with and

without patellofemoral pain across activities with progressively greater task

demands." Clin Biomech (Bristol, Avon) 23(2): 203-211.

Willy, R. W., K. T. Manal, et al. (2012). "Are mechanics different between male and

female runners with patellofemoral pain?" Med Sci Sports Exerc 44(11): 2165-

2171.

Yeung, S. S., E. W. Yeung, et al. (2011). "Interventions for preventing lower limb soft-

tissue running injuries." Cochrane Database Syst Rev(7): CD001256.

Zadpoor, A. A. and A. A. Nikooyan (2011). "The relationship between lower-extremity

stress fractures and the ground reaction force: a systematic review." Clin Biomech

(Bristol, Avon) 26(1): 23-28.

216

Zazulak, B., T. Hewett, et al. (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.

Zazulak, B. T., T. E. Hewett, et al. (2007). "The effects of core proprioception on knee

injury: a prospective biomechanical-epidemiological study." Am J Sports Med

35(3): 368-373.

Zhou, S. H., I. D. McCarthy, et al. (2000). "Geometrical dimensions of the lower lumbar

vertebrae--analysis of data from digitised CT images." Eur Spine J 9(3): 242-248.

Zimmerman, N. B., D. G. Smith, et al. (1988). "Mechanical disruption of human patellar

cartilage by repetitive loading in vitro." Clin Orthop Relat Res(229): 302-307.

217