UNIVERSITY OF CALGARY. Prescription of Specialized Footwear for Individuals with Knee Osteoarthritis. Ryan Tomas Lewinson A THESIS

Transcription

1 UNIVERSITY OF CALGARY Prescription of Specialized Footwear for Individuals with Knee Osteoarthritis by Ryan Tomas Lewinson A THESIS SUBMITTED TO THE FACULTY OF GRADUATE STUDIES IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY GRADUATE PROGRAM IN BIOMEDICAL ENGINEERING CALGARY, ALBERTA JANUARY, 2016 Ryan T. Lewinson 2016

2 ABSTRACT Wedged footwear insoles can reduce peak knee adduction moments during gait, which are associated with knee osteoarthritis development and progression; however, randomized trials have been mixed in terms of wedged insole clinical efficacy. To address methodological and technical limitations of past studies, the purposes of this thesis were to (1) identify the most suitable footwear control condition, from a biomechanical perspective, (2) identify a method that can predict the effect of wedged insoles on resultant knee adduction moments, and (3) evaluate the effects of a reduced knee adduction moment intervention on clinical outcomes for individuals with knee osteoarthritis. In fifteen healthy individuals, three sham footwear conditions were compared against participant s own footwear to determine if sham footwear were biomechanically inert. It was found that all three sham footwear conditions significantly altered biomechanical variables during gait, including the knee adduction moment, and thus participant s own footwear was recommended as the best control condition, from a biomechanical perspective, for future clinical studies. In fifteen healthy and nineteen knee osteoarthritis participants, a method was discovered that utilizes two dimensional video data from a single stepping motion to predict the change in knee adduction moment induced by wedged insoles during walking. When compared to actual walking data, the method successfully identified the correct insole intervention for 12/15 healthy and 17/19 knee osteoarthritis individuals. It is hoped this method may be implemented into clinical settings for improved footwear prescription capability for individuals with knee osteoarthritis. ii

3 Finally, a three month randomized trial was conducted. Biomechanical and clinical data were collected from 48 individuals with knee osteoarthritis at baseline, and participants were randomized to either an insole (knee adduction moment reduction) group, or a waitlist control group. At follow-up, no significant differences were noted between groups in terms of change in pain. Knee adduction moment reduction was not associated with reduced pain for the insole group. These data suggest that reduction of knee adduction moments do not confer a clinical benefit in the short term, but do not rule out the possibility that reduced moments are beneficial from a prevention standpoint, or for long term management. iii

4 PREFACE Three chapters of this thesis are based on manuscripts that have been either submitted to a scientific journal, or in the process of being submitted to a journal: Chapter 3 Lewinson RT, Worobets JT, Stefanyshyn DJ. (Submitted). Control conditions for footwear insole and orthotic research. Chapter 4 Lewinson RT, Stefanyshyn DJ. (Submitted). A method for predicting biomechanical response to wedged insoles. (Also filed as U.S. Provisional Patent No. 62/183,055 filed June 22, 2015). Chapter 5 Lewinson RT, Vallerand IA, Collins KM, Wiley JP, Lun VMY, Patel C, Woodhouse LJ, Reimer RA, Worobets JT, Herzog W, Stefanyshyn DJ. (In Preparation). Reduced knee adduction moments using wedged insoles for management of knee osteoarthritis: a 3-month randomized controlled trial. All chapters and subchapters were written in a manuscript-based style. Thus, some chapters may contain redundant information, mainly in the introduction and the methods sections since rationales and methods of the studies had some overlap. iv

5 ACKNOWLEDGEMENTS To my supervisors, Dr. Darren J. Stefanyshyn and Dr. Jay T. Worobets, I thank you greatly for providing me with the opportunity to study under your guidance, and for facilitating my professional and personal development. To my supervisory committee members, Dr. Walter Herzog and Dr. J. Preston Wiley, thank you for your commitment to this thesis, my research training, and for your invaluable scientific and clinical mentorship over the years. To Dr. Kevin A. Hildebrand, Dr. Tannin A. Schmidt, and Dr. Stephen P. Messier, thank you for serving on my examining committee and for your mentorship and feedback on this research. To all who have funded my work through scholarships, awards, and in-kind support, such as adidas International, Alberta Innovates Health Solutions, Association of Professional Engineers and Geoscientists of Alberta, Canadian Institutes of Health Research, Killam Trusts, Natural Sciences and Engineering Research Council of Canada, New Balance, and the University of Calgary, thank you for your generosity and support. To members of the Stefanyshyn research group, other students in the Human Performance Laboratory, and support staff in the Human Performance Laboratory, thank you for your availability to discuss and critique new research ideas, help in teaching and troubleshooting equipment, and for providing a terrific working environment. To staff at the University of Calgary Sports Medicine Centre, thank you for diagnosing patients, reviewing x-rays, and always being available to answer clinical questions related to this work. v

6 To Don McSwiney, thank you for your continuous support in media relations to promote new results and recruit new participants. Without your help, this thesis, and many of my other research projects, would have truly been impossible. To all study participants who generously volunteered their time and their knees for this thesis, it will always be appreciated. To my wife, Isabelle, my parents, Tom and Sharon, my siblings, Rebecca, Eric and Shaun, my in-laws, Andrew, Lise and James, my grandparents Anatol and Angelina, and Lyon and Doreen, aunts, uncles, cousins, and my good friends, thank you for supporting and motivating me throughout my training I would not be where I am today without you. vi

7 TABLE OF CONTENTS ABSTRACT... ii PREFACE... iv ACKNOWLEDGEMENTS...v TABLE OF CONTENTS... vii LIST OF TABLES... ix LIST OF FIGURES... xi LIST OF SYMBOLS, ABBREVIATIONS, NOMENCLATURES... xiii EPIGRAPH...xv CHAPTER ONE: INTRODUCTION Thesis Rationale Research Objectives Thesis Organization...4 CHAPTER TWO: REVIEW OF RELEVANT LITERATURE Epidemiology of Knee Osteoarthritis Clinical Presentation of Knee Osteoarthritis Biomechanics of Knee Osteoarthritis Calculation of Knee Joint Moments Biomechanics of Wedged Insoles Clinical Trials on Wedged Insoles for Knee Osteoarthritis Overview of Alternative Conservative Management Strategies Summary...28 CHAPTER THREE: CONTROL CONDITIONS FOR FOOTWEAR INSOLE AND ORTHOTIC RESEARCH Introduction Methods Participants Footwear Conditions Protocol Data Processing Statistical Analysis Results Discussion Conclusion...46 CHAPTER FOUR: PREDICTION OF KNEE JOINT MOMENT CHANGES DURING WALKING IN RESPONSE TO WEDGED INSOLE INTERVENTIONS Introduction Methods Participants Data Collection...49 vii

8 4.2.3 Data Processing Statistical Analysis Results Discussion Conclusion...64 CHAPTER FIVE: REDUCED KNEE ADDUCTION MOMENTS USING WEDGED INSOLES FOR MANAGEMENT OF MEDIAL KNEE OSTEOARTHRITIS: A 3- MONTH RANDOMIZED CONTROLLED TRIAL Introduction Methods Study Design Footwear Procedures & Measurements Sample Size Calculations Statistical Analysis Results Demographics Footwear & Biomechanics Follow-up Outcomes Co-interventions, Adherence & Adverse Events Discussion Conclusion...92 CHAPTER SIX: SUMMARY Overview of Rationale Results, Interpretations and Limitations Chapter 3 Control Conditions Chapter 4 Wedge Prediction Chapter 5 Three-Month Randomized Trial of Reduced KAMs Future Directions...98 REFERENCES APPENDIX A: BASELINE DATA COLLECTION FORM APPENDIX B: CHAPTER FIVE RCT PROTOCOL APPENDIX C: BASELINE & FOLLOW-UP PARTICIPANT SURVEY APPENDIX D: NEW ENGLAND JOURNAL OF MEDICINE PERMISSIONS APPENDIX E: BMC MUSCULOSKELETAL DISORDERS PERMISSIONS viii

9 LIST OF TABLES Table 2.1. Table 3.1. Table 3.2. Table 4.1. Table 5.1. Table 5.2. Results from studies evaluating the influence of laterally wedged footwear on reducing the peak knee adduction moment. All studies used a full-length wedge, and reported the first peak knee adduction moment. While many other studies have evaluated the influence of laterally wedged footwear on knee adduction moments in patients with knee OA, most studies do not report subject-specific results and so were not included in the table. Number of participants experiencing biomechanical change greater than ±10% relative to OS condition. X 2 values and p-values are also shown. P- values less than 0.05 indicate the proportion of participants experiencing changes greater than ±10% was significantly greater than the accepted proportion of 20%. OS represents the participant s own shoe, OSF represents the participant s own shoe with a flat 3 mm insole, SS represents a standardized shoe, and SSF represents a standardized shoe with a flat insole. Mean (S.D.) data across participants are shown for each variable within each footwear and movement condition. OS represents the participant s own shoe, OSF represents the participant s own shoe with a flat 3 mm insole, SS represents a standardized shoe, and SSF represents a standardized shoe with a flat insole. Slopes (ß1) and intercepts, with 95% confidence intervals shown in brackets, are shown for each of the lines of best fit produced for the simulated predicted KAM changes during walking vs actual KAM changes during walking, that are shown in Figure 4.3. Baseline characteristics for the two study groups, as well as the group of participants excluded following biomechanical testing are shown as number (%) unless denoted by *, where values are means (SD). P-values are shown for across group comparisons from Chi-square (X 2 ) for proportion data, and one-way ANOVA for continuous data. Biomechanics and footwear comfort data are shown as means (SD) between groups for the usual footwear condition, and also within the wedged insole group, where usual footwear is compared to the intervention insole. Mean differences (95% confidence interval) are shown for the wedged insole group, where negative values indicate a lower wedged insole result vs usual footwear. ix

10 Table 5.3. Table 5.4. Table 5.5. Clinical outcomes data at baseline and 3 months follow-up. Mean differences within groups are based on complete cases only (n=15 wedge group, n=18 control group) as per the statistical approach described in section 5.2.5, and thus the difference observed between the baseline and three months columns may not equate to those in the mean differences column. Co-intervention use and adherence characteristics are shown for both study groups. The new injuries reported throughout the 3 month study are shown as total number of participants experiencing at least one new injury. The specific injuries experienced are documented for each group; however, since some participants experienced more than one side effect, totals for the specific side effects do not equate to the total number of participants experiencing a side effect. x

11 LIST OF FIGURES Figure 2.1. Left, a schematic diagram of a right knee afflicted with medial knee osteoarthritis. Right, a radiograph of a left knee afflicted with medial knee osteoarthritis. Reproduced with permission from Felson 2006, New Engl J Med. Copyright Massachusetts Medical Society. Figure 2.2. An illustration showing the external knee adduction moment during the stance phase of walking in the right limb. Figure 2.3. On the right, a free body diagram of the foot and leg segments during the stance phase of walking is shown. Variables are color-coded based on the coordinate system used to originally measure or determine the data all variables are eventually converted to a joint coordinate system for calculation of resultant joint forces and moments. Only frontal plane variables are depicted in this figure; moments and forces acting in other planes have been excluded for visual clarity. On the left, the coordinate system conventions are shown by color code, as well as segment landmarks of importance for inverse dynamics calculations. Figure 3.1. (A) a superior view of a right and left 3 mm flat insole, (B) a lateral view of the left 3 mm insole, (C) a lateral view of the standardized shoe, and (D) an inferior view of the standardized shoe. Figure 3.2. Percent changes relative to OS during walking for the (A) peak external knee adduction moment, (B) external knee adduction angular impulse, (C) peak internal ankle inversion moment, and (D) maximum vertical loading rate. Figure 3.3. Percent changes relative to OS during running for the (A) peak external knee adduction moment, (B) external knee adduction angular impulse, (C) peak internal ankle inversion moment, and (D) maximum vertical loading rate. Figure 4.1. The single step procedure is shown. (1) The participant aligns themselves in a neutral position in front of the force plate. (2) The participant takes a step at a self-selected speed over the force plate, ensuring to land on the force plate with the limb of interest. (3) The participant steps off the plate and re-aligns to neutral stance on the opposite side of the force plate. In (2), when the vertical ground reaction force is at its maximum in the first half of stance phase, the mediolateral positions of the knee joint center, leg center of mass, ankle joint center, and foot center of mass are extracted. In the figure above, the sagittal plane is shown. Figure 4.2. Flow chart showing the data collection and analysis algorithm. xi

12 Figure 4.3. Relationship between predicted knee adduction moment (KAM) changes generated from the simulation analysis and actual changes to the knee adduction moment as measured during walking. The predicted r-squared value (r 2 p) is shown with lines of best fit. Details on the lines of best fit are shown in Table 4.1. The dotted orange lines show the 95% confidence interval of the line of best fit, and the dotted blue lines show the 95% prediction interval. Figure 5.1. Participant flow through the study. Figure 5.2. A frontal-plane diagram is shown for each of the static alignment variables evaluated. Details on the method of obtaining joint center positions are available elsewhere. Figure 5.3. Relationship between change in KOOS pain over 3 months and (A) change in KAM, (B) change in KAAI, and (C) change in 3D resultant moment are shown. xii

13 LIST OF SYMBOLS, ABBREVIATIONS, NOMENCLATURES Clinical BMI DXA KL KOOS OA PASE UCLA VAS Body mass index Dual x-ray absorptiometry Kellgren-Lawrence grade Knee osteoarthritis outcome score Osteoarthritis Physical activity scale for the elderly University of California at Los Angeles physical activity scale Visual analogue scale Biomechanical Note, some of the below variables are expressed relative to certain axes in thesis text. αf αl af al cmf cml FA FF FGRF FK FL IF IL KAAI KAM mf ml MA MGRF MK ra rgrf rla rlk ωf ωl 2D 3D Angular acceleration of the foot segment Angular acceleration of the leg segment Linear acceleration of foot center of mass Linear acceleration of leg center of mass Foot center of mass Leg center of mass Resultant internal force at ankle joint Weight force of foot segment Ground reaction force vector Resultant internal force at knee joint Weight force of leg segment Mass moment of inertia at foot center of mass Mass moment of inertia at leg center of mass Knee adduction angular impulse Knee adduction moment 1 st Peak Mass of foot segment Mass of leg segment Resultant moment at ankle joint Ground reaction free moment about vertical axis Resultant moment at knee joint Position vector from foot center of mass to point of FA application Position vector from foot center of mass to point of FGRF application Position vector from leg center of mass to point of FA application Position vector from leg center of mass to point of FK application Angular velocity of foot segment Angular velocity of leg segment Two-dimensional Three-dimensional xiii

14 Statistical α Significance level ANCOVA Analysis of covariance ANOVA Analysis of variance MANOVA Multivariate analysis of variance p p-value r Pearson s correlation coefficient r 2, R 2 Regression, Multiple regression coefficient SD Standard deviation X 2 Chi-square statistic 95% CI 95% confidence interval xiv

15 EPIGRAPH Specialization starts with the feet. Bruce L. Smith xv

16 CHAPTER ONE: INTRODUCTION 1.1 Thesis Rationale Osteoarthritis (OA) is a debilitating condition characterized by cartilage breakdown, eventually culminating in loss of physical function and disability. 1-4 Currently, there is no cure for OA instead, efforts have been focused towards conservative management strategies to prevent or delay the need for surgical joint replacement. 5 OA most frequently affects the knee, often the medial tibiofemoral compartment of the knee, where it is termed medial knee OA. 3 From a biomechanical perspective, development and progression of medial knee OA has often been attributed to an elevated peak external knee adduction moment (KAM) during the single-leg support phase of walking, 6 which is believed to be a good surrogate measure for actual tibiofemoral loading. 7 In fact, KAM magnitude has been positively correlated with pain, 8 as well as radiographic severity. 9 Consequently, reduction of KAMs during walking has been one of the primary knee OA research focuses for many years, as it is believed that this would reduce symptoms, and slow disease progression, potentially preventing or delaying the need for surgical knee replacement. One of the most commonly utilized conservative management strategies for medial knee OA has been a laterally wedged insole or orthotic. In most study designs, these insoles are given universally to all study participants in the experimental group, likely because these insoles have been shown to, on average across individuals, reduce KAMs during 1

17 gait However their clinical effectiveness remains inconclusive across studies, likely for a number of reasons. The first issue that could explain why mixed clinical results are seen in lateral wedge studies is that there is often disconnect between the intervention (lateral wedge), and the mechanical response (reduced KAMs). For instance, while lateral wedges do reduce KAMs on average across a group, there are typically 12-33% of individuals who actually experience increased KAMs during gait Despite this fact, most lateral wedge studies provide a lateral wedge to all participants in the experimental group without evaluating its biomechanical effectiveness, 13,21,22 and simply assume a reduced KAM was achieved. The consequence of this is that perhaps 12-33% of individuals receiving a lateral wedge insole actually experience clinical worsening, and therefore negatively bias the clinical results observed. Thus, the actual effect of reduced KAMs on clinical outcomes for individuals with medial knee OA is not presently known, as this has not been specifically controlled for. To fully answer this question, a clinical trial will need to ensure that all individuals receiving a wedged intervention actually experience reduced KAMs. A second possibility as to why mixed results are often observed regarding the clinical effectiveness of lateral wedges for individuals with medial knee OA is that in many clinical trials, a flat, supposedly biomechanically inert sham insole, or alternate control footwear is often given to the control group. 13,15 In fact, a recent systematic review has suggested that all future trials should utilize a flat insole control group, under the premise that this would help protect against placebo effects in clinical trials. 14 Unfortunately, this control condition has become standard practice without considering the possibility that flat insoles are in fact not biomechanically inert, and may induce changes to KAM magnitude 2

18 such that participants in the control arm of the study experience unintended clinical changes. 23 This scenario would be analogous to a drug trial where the placebo sugar pill induces physiological responses similar to the active drug under study in the experimental group certainly, this would render the effectiveness of the active intervention very challenging to interpret. Consequently, rigorous study of possible footwear control conditions is needed to inform what the best control condition may be for a clinical trial. The previous two paragraphs describe steps that are necessary to conduct a clinical trial aimed at understanding the effectiveness of reduced KAMs on clinical outcomes for individuals with knee OA. However, even if a clinical trial was conducted where the above two considerations have been made, an important research-clinical practice gap would still exists that would prevent the results from the clinical trial to be translated into medical practice. Specifically, how do we know which patients will benefit from a wedged insole? While in research settings it is possible to conduct biomechanical analyses on each individual patient entering the study to know if a given insole or orthotic is successful at reducing KAMs, such equipment, time, financial and technical resources are often not available in clinical settings. Currently, the only approaches that exist for predicting the influence of a wedged insole on KAM magnitude utilize full 3D kinematic and kinetic data collection, involve sophisticated and time consuming analysis procedures, and the predictive value of these approaches is quite low. 12,16 Thus, if a method could be developed that allowed for prediction of KAM response to a given insole or orthotic with relatively minimal, simple and inexpensive data collection and analysis, this could be incorporated into medical clinics to optimize management of knee OA. 3

19 1.2 Research Objectives There were three primary research objectives that this thesis sought to investigate. These objectives were as follows: (1) Identify the footwear/insole/orthotic condition that should be utilized as a control condition in randomized controlled trials aiming to assess the effects of reduced KAMs on clinical outcomes. (2) Identify a method that can be used to predict the expected change in KAM magnitude from a wedged insole intervention. (3) Identify the influence of reduced KAMs on clinical outcomes for individuals with medial knee OA. 1.3 Thesis Organization The remainder of the thesis has been organized into separate chapters that address each individual research objective. Firstly, an overview of relevant literature is provided in Chapter 2, which includes further information on the epidemiology, pathology and clinical presentation of OA, biomechanics of knee OA, biomechanics of wedged insoles, prediction of KAM magnitudes, and summary of existing clinical trials on reduced KAMs/wedged insoles for medial knee OA. Objective 1 is studied in Chapter 3, where four different footwear/insole control conditions are assessed from a biomechanical perspective. Objective 2 is addressed in Chapter 4, where a method for predicting the KAM response from wedged insole interventions is described. Objective 3 is answered in Chapter 5, where a three-month randomized controlled trial was conducted to understand the influence of reduced KAMs on clinical outcomes for individuals with medial knee OA. In the 4

20 randomized controlled trial, the results from Chapter 3 were incorporated for the control group, and only participants who experienced a KAM reduction in the experimental group were eligible for further participation. Chapter 6 presents a conclusion and interpretation of all results obtained from this thesis, as well as discussion of possible limitations and future directions. 5

21 CHAPTER TWO: REVIEW OF RELEVANT LITERATURE 2.1 Epidemiology of Knee Osteoarthritis Arthritis is one of the most prevalent musculoskeletal disorders, currently affecting 15% of people aged 15 years and older (~4.6 million Canadians), with an economic cost of $33 billion annually in Canada. 1,2 Most commonly, arthritis will take form as osteoarthritis (OA). Of all types of OA, the Centers for Disease Control of the United States has indicated that knee OA is the most prevalent, affecting 10% of adults over the age of Based on this statistic and population estimates from the 2011 Canadian Census, approximately 2.5 million Canadians currently have knee OA. Knee OA can take 4 forms: (1) medial tibiofemoral OA, whereby the OA is localized to the medial aspect of the knee joint, (2) lateral tibiofemoral OA, whereby the OA is localized to the lateral aspect of the knee joint, (3) patellofemoral OA, whereby the OA is localized to the patellofemoral joint, or (4) any combination of the previous three. Tibiofemoral disease is most frequently encountered in combination with patellofemoral disease (about 40% of all OA cases). 24 Between tibiofemoral compartments, medial compartment disease is about twice as prevalent as lateral compartment disease, 25 making medial knee osteoarthritis one of the most frequently encountered phenotypes. Due to this high rate of medial compartment disease, the following sections of this chapter, and remainder of this thesis are primarily focused on medial tibiofemoral OA, and all use of the terms knee OA, or medial knee OA refer to medial tibiofemoral OA unless otherwise specified. 6

22 While few large-scale epidemiologic studies have been performed as to factors associated with the development of medial knee OA, the available literature suggests that knee joint injury, obesity, and genetic factors may increase risk of OA development Clinical Presentation of Knee Osteoarthritis Knee OA is characterized by a number of physiological and structural changes. For instance, articular cartilage degeneration, decreased articular cartilage thickness, decreased joint space, increased thickness of subchondral bone, development of osteophytes (bony outgrowths), changes to synovial fluid composition, and swelling are all common with knee OA. 5 In some cases, symptoms will be severe enough to cause long-term disability. 2 Figure 2.1 shows a visual representation of a knee with typical characteristics of medial knee OA. 7

23 Figure 2.1. Left, a schematic diagram of a right knee afflicted with medial knee osteoarthritis. Right, a radiograph of a left knee afflicted with medial knee osteoarthritis. Reproduced with permission from Felson 2006, New Engl J Med. 4 Copyright Massachusetts Medical Society. 8

24 Clinically, patients with knee OA often present with knee joint pain, stiffness, and reduced physical function of the knee. According to the American College of Rheumatology radiographic and clinical criteria for diagnosis of knee OA, 29 a patient must present with (a) knee pain, (b) at least one of age over 50 years, morning stiffness lasting less than 30 minutes, or crepitus on active motion (a grating sound or sensation during movement), and (c) osteophytes on x-ray. Depending on a combination of factors, including (i) radiographic disease severity, which can be assessed by a radiologist on a numeric rating scale termed the Kellgren-Lawrence grade, (ii) patient s symptom severity, and (iii) patient s age, the patient will either undergo conservative management to slow the progression of the OA, or undergo knee replacement surgery. Often, the onset of OA is described as either post-traumatic, where a known injury preceded development of OA, or as idiopathic, where no identifiable trigger for OA is known, but may include a combination of genetics, previous heavy loading, comorbidities or inflammation. 2.3 Biomechanics of Knee Osteoarthritis It is widely believed that mechanical loading plays an important role in the progression and development of knee OA. Specifically, it is believed that increased stress on cartilage, whether from shear stresses, or compressive stresses, progressively induces damage until failure Commonly, during locomotion this loading is described in terms of the external knee adduction moment (KAM) that develops during the stance phase of walking (Figure 2.2). 6 This variable represents the resultant moment that occurs in the frontal-plane at the knee during gait. Theoretically, increased KAMs produce increased mechanical stress at the medial tibiofemoral articulation, and recently this has been 9

25 supported using finite element computer models, 34 as well as in in vivo experiments using instrumented knee prostheses. 35,36 Cross sectional studies have shown that increased KAMs are related to increased Kellgren-Lawrence OA severity grade, 9 and a prospective study has shown that the integral of KAMs with respect to stance time (i.e. KAM angular impulse) is associated with greater medial tibial cartilage volume loss over 12 months. 37 Additionally, it has been reported that KAM magnitude is associated with pain severity, as well as knee joint space narrowing over 6 years. 8 10

26 Figure 2.2. An illustration showing the external knee adduction moment during the stance phase of walking in the right limb. 11

27 There are a number of physiological and joint alignment parameters that appear to be closely related to the KAM. From an alignment perspective, a large research focus has been placed on frontal-plane knee alignment. In static terms, this is in reference to either genu varum or genu valgum. Genu varum refers to bow-leggedness, where the leg is adducted relative to the thigh, causing the knees to be separated when standing straight with feet together. Genu valgum refers to knocked-knees, where the leg is abducted relative to the thigh, and the patient may be unable to stand straight with feet together as the knees contact each other first. A similar concept is the quadriceps angle, or Q-angle. The Q-angle is measured as the angle between two lines: one drawn from the anterior superior iliac spine to the patella, and a second drawn from the patella to the tibial tuberosity. Thus, a larger Q-angle typically indicates greater genu valgum. In a study on runners with patellofemoral pain syndrome, it was found that individuals with larger static Q-angles typically had lower KAM magnitudes. 38 Longitudinally, it has been shown that risk of medial knee OA progression is four times higher in individuals with varus alignment, whereas risk of lateral knee OA progression is five times higher in individuals with valgus alignment. 39 In a dynamic sense, the term varus thrust is often used in the osteoarthritis literature. This term refers to the situation where, during the stance phase of walking, the knee joint thrusts laterally, placing the knee into a varus position. Intuitively, it can be interpreted that this rapid frontal-plane motion would place compressive stress on the medial aspect of the knee, and indeed, a study on women with medial knee osteoarthritis found that the amount of dynamic leg adduction during walking was closely related to KAM magnitude. 40 Additionally, larger varus thrust magnitudes have been associated with more severe pain and symptoms

28 From a physiology perspective, muscle strength is often discussed in regards to knee osteoarthritis and KAM magnitudes. In animal models it has been shown that quadriceps weakness via Botox injection results in rapid onset of osteoarthritic symptoms in the knee; 42 however, a recent study assessing individuals with early unilateral knee OA has suggested that differences in muscle cross sectional areas and isometric strength between the OA limb and contralateral healthy limb were not apparent at baseline or at 2- year follow-up. 43 In this study, muscle cross sectional area declined by only ~1% over the two year period, potentially explaining some of the discrepancy with animal studies where very large strength changes were induced. In the context of KAMs, one randomized controlled trial has found that a lower extremity muscle strengthening program does not affect KAM magnitude. 44 Additionally, while some studies have suggested hip muscle weakness as a factor associated with knee OA, 45 studies on healthy and osteoarthritis populations have highlighted that hip muscle weakness seems to be an outcome of knee OA rather than a predisposing factor. 46,47 In addition to the KAM, some researchers have begun studying the knee adduction angular impulse (KAAI), which is calculated by integrating the KAM-time curve during stance. While the KAM represents the resultant loading in the frontal-plane at a given instant in time, the KAAI represents the cumulative resultant frontal plane loading during the single-leg support phase of walking. Thus, it is believed that the KAAI could provide further information regarding the total load sustained by the knee. 48 In fact, one study highlighted that KAAI magnitude was correlated with medial tibial cartilage volume loss over a 12 month period, 37 while another showed that KAAI magnitude was related to pain severity

29 2.4 Calculation of Knee Joint Moments Knee adduction moments are generally calculated using an inverse dynamics procedure. This method utilizes a Newton-Euler approach for calculation of moments at a given joint using limb kinematic data and ground reaction force data. 50 Typically, ground reaction force data will be collected in 3D, allowing for determination of vertical, anterior/posterior and medial/lateral forces during the stance phase of the movement. 51 These forces are collected and originally expressed based on the force plate right handed coordinate system. In this thesis, the convention in Figure 2.3 is used, where the positive y force (anterior force applied to plate) would be projected out of the page. Simultaneously, a multi-camera system (often a near-infrared light based system) will be used to track the positions of retroreflective markers that have been placed over the lower extremity. 51 The 3D positions of each marker are collected and expressed in the laboratory right handed coordinate system, shown in Figure 2.3, where the positive x axis (anterior motion) would be projected out of the page. To determine the six degrees of freedom of segment motion (i.e. three translational directions, and three rotational directions), a minimum of three markers per segment are required. A 3D rotation matrix is applied to the data to determine the rotations of the segment. 52 Additional markers are placed over joints to define the joint center locations. 51 From this, segment lengths, masses and moments of inertia are defined based on literature reference values. 53,54 The center of mass locations are used to describe overall translation of the segment. A joint coordinate system is then defined for each joint. 55,56 This right handed coordinate system is used to describe the kinetics at each joint. It is defined by first creating the positive x axis as the line from the proximal joint center to the distal joint center. The 14

30 positive z axis is defined as the line from the medial joint marker to the lateral joint marker. The floating positive y axis (which would project out of the page in Figure 2.3) is then defined as the cross product of the x and z axes. Thus the origin of the joint coordinate system is at the joint center of interest. Newton-Euler based inverse dynamics equations are then used to solve for internal resultant forces and moments about each axis, starting at the most distal segment, and progressing proximally. For instance, based on the free body diagrams shown in Figure 2.3, resultant moments and forces are first solved for the ankle joint and then these data are applied to the free body diagram of the leg segment to solve for the internal resultant moments and forces at the knee joint. While forces and segment motions are originally collected/determined in their respective force or laboratory coordinate systems, in inverse dynamics equations all forces, moments and segment kinematics are solved relative to joint coordinate systems. 15

31 Figure 2.3. On the right, a free body diagram of the foot and leg segments during the stance phase of walking is shown. Variables are color-coded based on the coordinate system used to originally measure or determine the data all variables are eventually converted to a joint coordinate system for calculation of resultant joint forces and moments. Only frontal plane variables are depicted in this figure; moments and forces acting in other planes have been excluded for visual clarity. On the left, the coordinate system conventions are shown by color code, as well as segment landmarks of importance for inverse dynamics calculations. 16

32 In the case of solving resultant knee moments in Figure 2.3, firstly, resultant internal forces at the ankle joint would be solved using Newton s second law as a Fx F Ax F GRFx F Fx m F [ a Fy ] = [ F Ay ] + [ a Fz F Az F GRFy F GRFz ] + [ F Fy F Fz ] (1) where mf is the mass of the foot segment, af is the linear acceleration of the foot center of mass (in either the x, y or z direction), FGRF is the ground reaction force (in either the x, y or z direction), FF is the weight force of the foot (in either x, y or z direction) and FA is the internal ankle resultant force (in either x, y or z direction). The only unknown is FA, and thus the equations can be rearranged to solve for FA. With the resultant internal ankle forces now established, internal moments can be calculated about the foot segment center of mass in the ankle joint coordinate system as [ M Fcmx M Fcmy M Fcmz ] = [ I Fxx I Fyy I Fzz ] [ α Fx α Fy 0 ω Fz ω Fy ] + [ ω Fz 0 ω Fx ] [ α Fz ω Fy ω Fx 0 I Fxx I Fyy I Fzz ] [ ω Fx ω Fy ω F ] r Ax F Ax r GRFx F GRFx M Ax M GRFx = [ r Ay ] [ F Ay ] + [ r GRFy ] [ r Az F Az r GRFz F GRFy F GRFz ] + [ M Ay M Az ] + [ M GRFy M GRFz ] (2) 17

33 where MFcm are the sum of moments about the center of mass, IF are the principal moments of inertia about the foot, αf are the angular accelerations of the foot, ωf are the angular velocities of the foot, FGRF are the external ground reaction forces, rgrf are the displacement vectors from the foot center of mass to the point of application of the ground reaction force, FA are the resultant forces at the ankle joint, ra are the displacement vectors from the foot center of mass to the point of application (ankle joint center) of the ankle resultant force, MGRF is the ground reaction free moment, which is only a non-zero quantity about the force plate vertical axis, and MA are the internal resultant ankle moments. This equation can be rearranged and solved for MA. Applying Newton s third law to the leg segment, the resultant forces and moments found at the ankle are applied in equal magnitude but opposite direction for the leg segment. These are then transformed into the new knee joint coordinate system. From this, resultant forces at the knee can then be calculated as a Lx F Kx F Ax F Lx m L [ a Ly ] = [ F Ky ] + [ F Ay ] + [ a Lz F Kz F Az F Ly F Lz ] (3) where ml is the mass of the leg segment, al is the linear acceleration of the leg center of mass (in either x, y or z direction), FA is the internal ankle reaction force (in either x, y or z direction), FL is the weight force of the leg (in either x, y or z direction) and FK is the internal knee resultant force (in either x, y or z direction). The only unknown is FK, and thus the 18

34 equation can be rearranged to solve for FK. With the resultant internal knee forces now established, internal moments can be calculated about the leg segment center of mass as [ M Lcmx M Lcmy M Lcmz ] = [ I Lxx I Lyy I Lzz ] [ α Lx α Ly 0 ω Lz ω Ly I Lxx 0 0 ω Lx ω Ly ] + [ ω Lz 0 ω Lx ] [ 0 I Lyy 0 ] [ ] α Lz ω Ly ω Lx I Lzz ω Lz r LKx F Kx r LAx F Ax M Ax M Kx = [ r LKy ] [ F Ky ] + [ r LAy ] [ F Ay ] + [ M Ay ] + [ r LKz F Kz r LAz F Az M Az M Ky M Kz ] (4) where MLcm are the sum of moments about the leg center of mass, IL are the principal moments of inertia about the leg, αl are the angular acceleration of the leg, ωl are the angular velocities of the leg, FA are the internal ankle reaction forces, rla are the displacement vectors from the leg center of mass to the point of application (ankle joint center) of the ankle internal reaction forces, FK are the resultant internal forces at the knee, rlk are the displacement vectors from the leg center of mass to the point of application (knee joint center) of the internal knee resultant forces, MA are the internal resultant ankle moments, and MK are the internal resultant knee joint moments. This equation can be rearranged and solved for MK, where the y (frontal-plane) component is the KAM. It must be stressed that the above calculation yields internal resultant moments, or moments produced as a result of tissues and forces occurring within the body. It is deemed a resultant as it is the cumulative effect of all smaller forces and moments acting on the joint and thus a limitation of this approach is that one cannot know precisely what caused 19

35 these forces and moments (i.e. the contribution from ligaments, muscles, contact forces etc.) without further mathematical or musculoskeletal modeling techniques. 34,52,57 Typically during gait, one will experience a large internal knee abduction moment, 10,58,59 or MKy in equation 4. To convert this to an external knee adduction moment (KAM), researchers simply take the negative of the internal abduction moment. 36 As a result, the two terms can be used interchangeably, but refer to different interpretations. Specifically, the internal moment is due to a number of factors intrinsic to the body such as muscle or ligaments. The external moment refers to the typical movement experienced by the leg segment during stance phase, which is adduction. 60 This movement draws the medial tibiofemoral articulation into compression, and thus using the term external adduction moment has gained popularity in the context of medial knee osteoarthritis. From the KAM, the peak value (peak KAM), and the knee adduction angular impulse (KAAI) is often calculated. Conceptually, the KAAI can be understood as the area beneath the KAM-time curve (shown in Figure 2.2), and mathematically it is defined as t2 KAAI = KAM(t) dt t1 (5) where KAAI is the knee adduction angular impulse, dt is the instantaneous time, KAM is the knee adduction moment at time t, t1 is touchdown, and t2 is toe-off. While the above inverse dynamics approach is the most common for determination of KAMs, another approach exists whereby the frontal-plane ground reaction force vector is simply multiplied by its perpendicular distance to the knee joint center. 61,62 A recent 20

36 analysis by Lewinson and colleagues 62 has shown that this approach underestimates the magnitude of the KAM, but serves reasonably well in the context of approximating the change to the KAM induced by a footwear intervention. Despite this method s inability to correctly estimate KAM magnitude, researchers have held the notion that GRF magnitude and/or GRF-knee lever arm length are key factors in determining KAM magnitude. 6, Biomechanics of Wedged Insoles Wedged footwear has been defined in a number of ways: (1) footwear with a wedged outsole, 63 (2) a wedged orthotic or insole placed within the shoe, 19 or (3) a variable stiffness shoe that simulates a wedge by increasing shoe sole stiffness on either the medial or lateral edge. 15 While wedged orthotics/insoles tend to be the most commonly available for consumer purchase and in clinical practice, in all cases it has been shown that wedged footwear can significantly influence KAMs during locomotion. Specifically, on average across individuals, lateral wedges tend to reduce KAMs, while medial wedges tend to increase KAMs. 10,18,64-66 Typically, it is believed that these changes occur as a result of a medial or lateral shift to the center of pressure beneath the foot, which in turn changes the ground reaction force to knee joint center lever arm length, thereby altering KAM magnitude (Figure 2.2). Based on these average trends, lateral wedges have become the main wedge prescribed for management of knee OA. Indeed, the Osteoarthritis Research Society International (OARSI) clinical guidelines for management of knee OA have suggested using laterally wedged footwear for management of knee OA. 67 The problem with this is that about 23% of patients receiving a lateral wedge will actually experience increased KAMs that could potentially result in OA worsening (Table 2.1). 21

37 Table 2.1. Results from studies evaluating the influence of laterally wedged footwear on reducing the peak knee adduction moment. All studies used a full-length wedge, and reported the first peak knee adduction moment. While many other studies have evaluated the influence of laterally wedged footwear on knee adduction moments in patients with knee OA, most studies do not report subject-specific results and so were not included in the table. Study Wedge Angle Mean change to KAM # of subjects experiencing increased KAMs % of subjects experiencing increased KAMs Butler et al., o a 8.9% 3/20 15% Chapman et al., o 11.4% 23/70 33% Erhart et al., N/A b 3.5% 11/34 32% Hinman et al., o 5.4% 5/ % Hinman et al., o 12% 2/ % Kakihana et al., o 5.6% 2/ % Kakihana et al., o 5.5% 9/ % TOTAL/AVERAGE 6.1 o 8.9% 55/ % a Butler et al., used subject specific wedging. Their mean wedge angle was 9.6 o, with a standard deviation of 3.2 o ; b Erhart et al., used a variable stiffness shoe, where the sole was stiffer laterally to simulate a wedge. 22

38 Relatively few studies on medial wedges have been performed on individuals with knee osteoarthritis, and even fewer have reported proportions of individuals experiencing a KAM increase or reduction as a result of the wedge. Although precise values were not given, Jones et al. 68 have suggested that when individuals were given a lateral wedge with a medial arch support (which functions as a medial wedge), at least 54% experienced reduced KAMs the opposite of what is typically seen with medial wedges. In studies on healthy individuals during running, it has also been shown that 40% of individuals receiving a medial wedge may in fact experience reduced KAMs. 10,66 Therefore, from a biomechanical perspective, it remains a possibility that medial wedges could serve as an alternate intervention to provide to those individuals who do not respond favorably (i.e. reduced KAM) to a lateral wedge, or in the event that a medial wedge induces a greater KAM reduction than a lateral wedge. This insole type has not been well studied in individuals with medial knee osteoarthritis. One major limitation to the above notion of utilizing both medial and lateral wedges for individuals with knee OA, and the concept that a large proportion of individuals may not respond favorably to lateral wedges is that it is currently not possible to predict the type of wedge an individual will need to reduce KAMs. Instead, a full biomechanics data collection, as described in section 2.4 is required, which can be quite expensive, time consuming, and difficult to perform without a trained biomechanics professional. Some researchers have attempted to identify simple approaches to predicting the biomechanical response to wedged insoles. For instance, Hinman et al. 12 found that the change in ground reaction force to knee joint center lever arm length was significantly associated with KAM magnitude change; however, this only accounted for 46% of the variance in the data. 23

39 Additionally, Chapman et al. 16 suggested that the frontal-plane ankle angle was associated with KAM magnitude; however, again this was only a moderate relationship. Importantly, both of these approaches also require 3D data collection to quantify the predictive variables, and so offer no advantage in a clinical sense. For a predictive variable to be useful in a clinical sense for the prescription of wedged insoles, the variable must be easily quantifiable and analyzed at a low cost, in a low amount of time, and have high predictive ability. 2.6 Clinical Trials on Wedged Insoles for Knee Osteoarthritis While lateral wedges have been studied for many years for knee osteoarthritis, studies focusing on medial wedges and their clinical effectiveness are scarce. Recently, Parkes and colleagues 14 identified twelve studies of sufficiently high quality to be included in their meta-analyses on the effectiveness of wedged footwear for management of knee osteoarthritis. Treatment durations with wedged footwear have ranged from as short as two weeks, 69 to as long as 2 years. 21 Generally, follow-up times of 3 months (n=3) and 12 months (n=3) appear to be the most common, perhaps representing a short-term and long-term follow-up time. In the meta-analysis by Parkes et al., 14 there was no association between treatment effect size and treatment duration, suggesting that changes in pain and symptoms occur fairly quickly. Sample sizes have varied across trials, generally between individuals in each of the control and experimental groups, 15 but some of the larger studies have included upwards of 60 individuals per group. 13 It has been shown that larger studies tend to find 24

40 null effects of wedged footwear in terms of their effects on pain and symptoms, while smaller ones generally show either a clinical benefit, or a null result, but never a negative result. 14 Parkes et al. 14 reported that similar patient profiles were utilized across all the studies regardless of size, so more likely this small-study bias is a result of more focused, intensive interventions rather than due to inappropriate sample sizes. For instance, larger studies tend to assign all individuals the same intervention without monitoring of baseline biomechanics, 13 while smaller studies may personalize the assigned orthotic or add complementary interventions. As alluded to above, many studies assign the same intervention to all individuals in the study. As described in Section 2.5, this approach may result in a high proportion of individuals who experience a negative KAM response, and could result in those individuals experiencing symptom worsening. In fact, only one of the twelve trials identified by Parkes et al. 14 measured KAM changes at baseline, but still included those individuals who experienced KAM increases from the wedged footwear. 15 While these participants would be in the minority, this clinical worsening could bias the treatment group enough so as to mask the clinical effectiveness of the wedged footwear for those who did experience a KAM reduction. Consequently, the true effect of reduced KAMs on clinical outcomes for individuals with knee OA remains unknown. One of the primary findings in the meta-analysis by Parkes et al. 14 was that studies that used a flat, supposedly inert insole as a control condition tended to show null effects of wedged footwear, whereas those that utilized the participant s own shoe, without further modification, tended to show clinical benefit of wedged footwear. The authors proposed that this discrepancy was likely a result of placebo effects the scenario whereby receiving 25

41 any insole intervention induces a subjective perception of improvement. Specifically, they believed that in studies where the participants own shoe was used as the control condition, placebo effects alone may explain the clinical benefit observed in the insole group, whereas in studies with a flat control insole, both groups would have experienced placebo effects thus rendering comparisons between groups non-significant. While placebo effects are indeed an important aspect of any randomized trial, this can only be effectively addressed if the proposed control condition is indeed inert from a physiological or mechanical perspective. For instance, a sugar pill in drug trials would not serve as a good placebo control if it induced similar physiologic responses to the active drug in the experimental group. The assumption is that flat insoles are indeed biomechanically inert in terms of their effect on KAMs; however this has never been tested experimentally. One study by McCormick et al. 70, has shown that flat insoles alter plantar pressure distributions during walking, and since KAM magnitude is correlated with the center of pressure beneath the foot, 12 it remains a possibility that KAMs could be altered with flat insoles. The consequence of this could be that some individuals in the control group of studies utilizing a flat insole actually experienced altered KAM magnitudes that could affect clinical symptoms thus masking the effect of the experimental condition. Based on the likelihood of improper wedge prescription to the experimental group participants (i.e. some experiencing increased KAMs), as well as the possibility that some individuals in control groups utilizing flat insoles experienced actual clinical changes from an unintentional biomechanical intervention, it remains impossible to interpret the true effectiveness of wedged footwear and reduced KAMs for individuals with knee OA. 26

42 2.7 Overview of Alternative Conservative Management Strategies Aside from wedged footwear, a number of alternative conservative management therapies exist for medial knee osteoarthritis, most notably offloader knee braces, intraarticular injections of corticosteroids or hyaluronic acid, and physiotherapy. Offloader knee braces exert forces on the lateral aspect of the thigh and leg in attempt to correct varus alignment, and to reduce the KAM during gait. While these offloader braces can reduce KAMs, 71 it has been shown that lateral wedges may have a greater effect on KAM reduction, 72 and that many individuals report a high degree of discomfort with these braces due to their bulkiness, difficulty in achieving a good fit, and skin irritation. 73 Intra-articular knee injections generally take one of two forms. The first is a corticosteroid (cortisone) injection, which serves to reduce joint inflammation and pain. 5 Systematic reviews have shown cortisone injections to be effective in reducing short term pain, 74 but not in terms of functional improvement. 75 The second injection type is a viscosupplement called hyaluronic acid, which is a naturally occurring lubricant in the knee. This injection helps to lubricate joints and reduce pain. 76 Clinical effectiveness is currently mixed across trials, with some patients reporting good improvement, and some reporting symptom worsening and increased risk for adverse events. 77 In addition, one study has found that a hyaluronic acid injection resulted in an increased KAM during gait. 78 These authors proposed that while the injection may offer symptom relief in the short term, there is a possibility that they may also increase rate of overall OA disease progression. Physiotherapy for knee OA usually involves manual and/or home exercise intervention. Recent evidence suggests that the effects of these interventions usually do not affect KAM magnitude, but can offer symptom improvement in some cases. 44,79 Generally, 27

43 there is limited evidence in terms of the ideal physical therapy intervention for knee OA, and the clinical benefit may be small Summary Knee OA is a degenerative condition that appears to be characterized by increased resultant knee joint loading in the frontal-plane. Given the current state of mixed evidence for numerous conservative management strategies for knee OA, it seems to be of critical importance to begin to narrow trial scopes and identify personalized approaches to management of knee OA. The rationale for utilizing a wedged insole intervention for management of knee OA is compelling, as it offers a well-tolerated and highly noninvasive approach to OA management that directly targets loading in the knee. 28

44 CHAPTER THREE: CONTROL CONDITIONS FOR FOOTWEAR INSOLE AND ORTHOTIC RESEARCH 3.1 Introduction Footwear insoles or orthotics are commonly utilized as an intervention to prevent, treat or manage a variety of musculoskeletal disorders. Typically, the desired objective of the insole is to modify an individual s gait so as to alter kinetic variables that are believed to be injurious. For example, in osteoarthritis research, an investigator may attempt to reduce the peak external knee adduction moment during gait; 6,17,68 in patellofemoral pain research, it may be desirable to modify the external knee adduction (or internal knee abduction) angular impulse; 59 for ankle sprains, the ankle inversion moment might be reduced; 81 and for tibial stress syndrome, a reduced maximum vertical ground reaction force loading rate may be desired. 82 Each of these kinetic variables during gait can be modified by specialized footwear insoles or orthotics. 10,17,66,83 Clinically, the effectiveness of footwear insoles or orthotics generally show mixed results, where some trials show a clinical benefit and some trials do not. 14,84 In one recent systematic review on insoles for knee osteoarthritis, it was proposed that a potential reason as to why trials show differing clinical results is that different control conditions are often used across studies. 16 Specifically, in trials that used a flat insole as a control condition, no clinical benefit of specialized insoles or orthotics was observed; however, trials that used the participant s own shoe as a control condition did find clinical benefits. The authors assumed that flat insoles are biomechanically inert, in the way that a sugar pill might be chemically inert for a drug trial, and therefore speculated that the discrepancy in clinical 29

45 effectiveness between studies could be attributed to placebo effects in studies that did not use a flat control insole. While flat insoles and the participant s own shoe remain popular choices for control conditions in insole and orthotic research, researchers have also used a standardized shoe across participants, or a standardized shoe across participants with the addition of a flat insole. 10,68,83 Clearly, a number of options exist for a control condition, yet there does not appear to be a consensus on the most appropriate option for clinically-oriented biomechanics research. Recently, in response to the systematic review by Parkes et al., 14 an alternate explanation as to why clinical benefit depended on control condition was proposed: that the flat insole control condition was in fact not biomechanically inert. 23 Or in other words, it is possible that the flat insole conditions actually caused changes to participant s biomechanics such that it induced a clinical response in the control group, making the detection of differences between control and experimental groups more difficult. Surprisingly, very little research is available on footwear control conditions to support or refute this theory. In one study by McCormick et al., 70 it was shown that plantar pressure distributions beneath the foot are altered during gait with the addition of a flat control insole, lending support to the theory that perhaps the flat control insert and other control conditions alter the individual s biomechanics relative to what they normally experience with their own shoes. This is important, since the goal of a biomechanical control condition in clinical research should be to ensure the individual s biomechanics are unmodified relative to what they normally experience to prevent a clinical response from occurring. 30

46 The purpose of the present study was to conduct a comparison of commonly utilized control conditions in the footwear insole and orthotic literature to identify if kinetic variables at the ground, ankle and knee that are associated with musculoskeletal injury, are altered relative to what a participant would normally experience while wearing their own shoe. These footwear conditions included the participant s own shoe, the participant s own shoe with a flat insole, a standardized shoe, and a standardized shoe with a flat insole. It was hypothesized that peak external knee adduction moments, external knee adduction angular impulses, peak internal ankle inversion moments, and maximum vertical loading rates variables associated with musculoskeletal injury that are modifiable through insole interventions would be significantly altered in each of the proposed control conditions relative to the participant s own shoe. 3.2 Methods Participants Fifteen healthy individuals were recruited to participate in this study (8 male, 7 female, mean±sd age of 22±1.7 years, height of 176.4±10.7 cm, and mass of 73.6±11.5 kg). All participants were free from any lower extremity pain or injury at the time of data collection, and all participants gave written informed consent prior to data collection, in accordance with the University s health research ethics board Footwear Conditions Four footwear conditions were tested in this study. Since the goal of most footwear insole studies is to modify the participants biomechanics relative to what they normally 31

47 experience, the participants own shoe with no added footwear insole (OS) was considered to be the gold standard control condition (since it would not change the individual s normal biomechanics), and therefore served as the baseline condition in the present study. It should be emphasized that the participant s own shoe is not being proposed here as the universally optimal control shoe rather, as compared to other footwear control conditions that are often utilized in footwear clinical studies, the participant s own shoe is being proposed as the ideal control condition from a biomechanical perspective only. The three conditions that were tested against the baseline OS condition were (1) the participants own shoe with an added flat 3 mm insole (OSF), (2) an alternate, standardized shoe (adidas Mana) with no insole (SS), or (3) the standardized shoe with an added flat 3 mm insole (SSF). In the context of footwear insole studies, the OSF condition would represent the researcher attempting to control for placebo effects of an insole while not influencing the individual s biomechanics. The SS condition would represent the researcher attempting to eliminate bias that could arise from different shoe types across the participants, and the SSF condition represents the researcher attempting to control for varying footwear models, while also controlling for a placebo effect of insoles. Various flat 3 mm insoles were prepared using ethylene vinyl acetate (EVA) sheets, and cut to the shape of a footwear insole (fitted to Women s size US6 to US10 or Men s size US8 to US12). The insole was not modified further, and kept entirely flat in attempt to keep the insole as biomechanically inert as possible. The flat insole and standardized shoe are shown in Figure

48 A. B. 3 mm C. D. Figure 3.1. (A) a superior view of a right and left 3 mm flat insole, (B) a lateral view of the left 3 mm insole, (C) a lateral view of the standardized shoe, and (D) an inferior view of the standardized shoe. 33

49 3.2.3 Protocol Three retroreflective markers were secured to each participant s right leg, and an additional three markers were placed on the right OS shoe, and the right SS shoe. In a randomly assigned order, participants completed five successful trials walking at 1.6±0.08 m/s, and 5 successful trials running at 4.0±0.2 m/s with each of the four footwear conditions (OS, OSF, SS, SSF) while an eight camera Motion Analysis system (Motion Analysis Corp., Santa Rosa, CA) collected three dimensional marker trajectories at a frequency of 240 Hz, and a force platform (Kistler AG, Winterthur, Switzerland) mounted flush with the lab floor collected three dimensional ground reaction forces at a frequency of 2400 Hz. Successful trials were defined as those in which the participant landed near the center of the force plate with their right foot, did not touch the force plate with their left foot, and was within the target gait speed as monitored by two photocells. Participants were provided with a two minute acclimation period to each footwear condition prior to data collection. Immediately following data collection with each footwear condition, additional retroreflective markers were placed over the medial and lateral epicondyles of the right thigh, and the medial and lateral malleoli on the right leg. The participant then stood in the anatomical position with both feet on the force platform, with attempt to align the participants joints with the laboratory coordinate system. Static marker and force plate data were then collected. This was done for each footwear condition Data Processing Kinematic and kinetic data from dynamic and static trials were imported into KinTrak v7.0 (University of Calgary, Calgary, AB). Kinematic and kinetic data were 34

50 filtered using 4 th order low-pass Butterworth filters with cutoff frequencies of 12 Hz and 50 Hz, respectively. From the static trials, ankle and knee joint centers were defined as the midway point between the malleoli and epicondyle markers, respectively. 51 Leg segment lengths were defined as the length from the knee joint center to the ankle joint center, and the foot segment lengths were defined as the length from the ankle joint center to the lab floor. Segment centers of mass locations, segment masses, and segment moments of inertia were then determined from literature values. 53,54 These parameters, from each of the static trials, were then applied to their associated dynamic trials. A joint coordinate system was defined for the ankle and knee joints, 55,56 and inverse dynamics calculations were performed. 50 For each dynamic trial, the peak external knee adduction moment, external knee adduction angular impulse, peak internal ankle inversion moment and maximum vertical ground reaction force loading rate were extracted. These variables were selected as they have been associated with the development and/or progression of musculoskeletal injuries such as knee osteoarthritis, 6 patellofemoral pain syndrome, 59 ankle sprain, 81 and/or tibial stress syndrome Statistical Analysis All statistical analysis was performed using MATLAB R2015a (MathWorks, Cambridge, MA), at a significance level of Biomechanical variables were averaged across the five trials in each footwear-movement condition for each participant to obtain a single mean value, resulting in a mean value for each variable under each shoe condition for each participant. 35

51 In studies of wedged insoles, it is often observed that participants will experience biomechanical changes in directions that are opposite to group mean responses, 16,19 and it has been proposed that this has likely biased footwear trial outcomes. 23 Therefore, to account for the possibility that changes to the tested biomechanical variables may either increase relative to usual footwear, or decrease relative to usual footwear for different individuals, percent changes relative to the baseline OS condition were determined for each variable-movement combination in the OSF, SS and SSF conditions. The primary analysis was then performed as a one-sample Chi-Square (X 2 ) test for proportions (two-tailed) to compare the proportion of individuals who experienced a biomechanical change greater than ±10% with the OSF, SS and SSF conditions relative to the OS condition and those who experienced a change less than ±10%. The threshold for percent change of ±10% was chosen since, (a) for the knee adduction moment variable, this change represents a change that approximates the changes typically experienced by wedged insoles an intervention designed specifically to reduce knee adduction moments, 16,19 (b) a ±10% change is larger than what is typically reported for within day trial repeatability for joint kinetics variables, and (c) this threshold for biomechanical relevance has been used previously. 66 For each of the OSF, SS and SSF conditions to be deemed biomechanically inert, an accepted proportion of 20% (or 3/15) for those who experience changes greater than ±10% for each variable was selected. The accepted proportion of 20% was chosen since this percentage roughly coincides with the proportion of individuals for whom wedged insole interventions do not reduce knee adduction moments. 16,19,20 Thus, we considered that wedged insoles, and control conditions should have roughly the same failure rate, and that a failure rate beyond 20% for a control group would make conducting 36

52 a clinical trial rather challenging. Therefore, based on the mathematical formulation of the Chi-Square test, the test would return (a) no significant difference in proportions when the actual proportion was equal to or less than 40% (6/15), or (b) significant difference in proportions when the actual proportion was equal to or greater than 46.7% (7/15). Secondarily, repeated measures ANOVA were used to compare magnitudes of each variable across footwear conditions within each movement condition to assess if there was indeed a systematic change induced by the control conditions. If significant differences were detected, two-tailed paired-samples t-tests were utilized to identify these differences; however, only comparisons relative to the OS condition were made since the objective was to identify conditions that are biomechanically similar to the OS condition. 3.3 Results Percent changes to each biomechanical variable are shown for each participant during walking and running in Figures 3.2 and 3.3, respectively. As can be seen, the percent change directions and magnitudes differed dramatically across participants and footwear conditions. Table 3.1 shows the number of participants who experienced a percent change of ±10% with each footwear condition relative to the OS condition, and for each variable. As can be seen, the majority of footwear-variable-movement combinations resulted in at least 7/15 (46.7%) of participants experiencing a change of ±10%, resulting in significantly greater proportions than the accepted proportion of 20% (3/15). 37

53 Figure 3.2. Percent changes relative to OS during walking for the (A) peak external knee adduction moment, (B) external knee adduction angular impulse, (C) peak internal ankle inversion moment, and (D) maximum vertical loading rate. Figure 3.3. Percent changes relative to OS during running for the (A) peak external knee adduction moment, (B) external knee adduction angular impulse, (C) peak internal ankle inversion moment, and (D) maximum vertical loading rate. 38

54 Table 3.1. Number of participants experiencing biomechanical change greater than ±10% relative to OS condition. X 2 values and p-values are also shown. P-values less than 0.05 indicate the proportion of participants experiencing changes greater than ±10% was significantly greater than the accepted proportion of 20%. OS represents the participant s own shoe, OSF represents the participant s own shoe with a flat 3 mm insole, SS represents a standardized shoe, and SSF represents a standardized shoe with a flat insole. Walking (n=15) Variable OSF SS SSF External Knee Adduction Moment Peak 7 (X 2 =6.67, p=0.010) 10 (X 2 =20.4, p<0.001) 10 (X 2 =20.4, p<0.001) External Knee Adduction Angular Impulse 6 (X 2 =3.75, p=0.052) 8 (X 2 =10.4, p=0.001) 11 (X 2 =26.7, p<0.001) Internal Ankle Inversion Moment Peak 9 (X 2 =15.0, p<0.001) 13 (X 2 =41.7, p<0.001) 13 (X 2 =41.7, p<0.001) Vertical Loading Rate Maximum 8 (X 2 =10.4, p=0.001) 13 (X 2 =41.7, p<0.001) 12 (X 2 =33.8, p<0.001) Running (n=15) Variable OSF SS SSF External Knee Adduction Moment Peak 6 (X 2 =3.75, p=0.052) 13 (X 2 =41.7, p<0.001) 10 (X 2 =20.4, p<0.001) External Knee Adduction Angular Impulse 7 (X 2 =6.67, p=0.010) 9 (X 2 =15.0, p<0.001) 10 (X 2 =20.4, p<0.001) Internal Ankle Inversion Moment Peak 9 (X 2 =15.0, p<0.001) 11 (X 2 =26.7, p<0.001) 10 (X 2 =20.4, p<0.001) Vertical Loading Rate Maximum 6 ( X 2 =3.75, p=0.052) 10 (X 2 =20.4, p<0.001) 8 (X 2 =10.4, p=0.001) 39

55 Comparisons of mean data can be found in Table 3.2. For all variables tested while walking, no significant differences were found between footwear conditions. During running, no significant differences were detected for peak external knee adduction moments, external knee adduction angular impulses, or maximum vertical loading rate; however, significance was detected for the peak internal ankle inversion moment across conditions (p=0.024). When conducting post-hoc tests for this variable, no significant comparisons were found relative to the OS condition (OS vs OSF p=0.58; OS vs SS p=0.073; OS vs SSF p=0.85). 40

56 Table 3.2. Mean (S.D.) data across participants are shown for each variable within each footwear and movement condition. OS represents the participant s own shoe, OSF represents the participant s own shoe with a flat 3 mm insole, SS represents a standardized shoe, and SSF represents a standardized shoe with a flat insole. Walking Summary Data (n=15) Variable OS OSF SS SSF ANOVA p-value Peak external knee adduction moment [Nm] 47.4 (15.8) 46.0 (16.5) 46.9 (16.2) 45.9 (16.6) External knee adduction angular impulse [Nms] 15.8 (5.9) 14.9 (5.7) 16.4 (6.4) 15.4 (6.4) Peak internal ankle inversion moment [Nm] 11.6 (3.9) 11.8 (4.8) 10.1 (4.2) 11.4 (4.8) Peak vertical loading rate [N/s] 15,429 (4,584) 15,938 (4,725) 15,886 (2,779) 16,623 (2,759) Running Summary Data (n=15) Variable OS OSF SS SSF ANOVA p-value Peak external knee adduction moment [Nm] 87.7 (33.1) 82.9 (37.6) 83.5 (35.0) 87.0 (37.8) External knee adduction angular impulse [Nms] 10.3 (5.4) 10.1 (5.9) 10.1 (5.6) 10.4 (5.9) Peak internal ankle inversion moment [Nm] 27.3 (10.6) 28.3 (11.3) 22.8 (9.1) 26.8 (11.0) 0.024* Peak vertical loading rate [N/s] 64,216 (21,633) 60,337 (20,343) 55,369 (14,066) 54,923 (15,147) *ANOVA detected significant difference; no post-hoc tests found to be significant 41

57 3.4 Discussion The purpose of this study was to conduct a comparison of commonly utilized control conditions in the footwear insole and orthotic literature to identify if kinetic variables at the ground, ankle and knee are altered relative to what a participant would normally experience while wearing their own shoe. Specifically, the rationale for this study was to identify the most suitable control condition, from a biomechanical perspective, for use in future randomized controlled trials. It was hypothesized that biomechanics would be significantly altered in each of the conditions relative to the participant s own shoe. This hypothesis was supported since there was a significantly larger proportion of individuals than the expected proportion of 20% who experienced a biomechanical change of ±10% for many of the footwear conditions, movements and variables tested. However, the secondary analysis in this study found that these changes were not consistent across participants, in that some individuals experienced increased variable magnitudes and some experiencing decreased variable magnitudes. Therefore, while the directionality of change does not appear to be consistent across participants for each of the footwear conditions tested, a large proportion of participants did experience biomechanically important changes relative to what they experience in their own shoe. McCormick et al., 70 found that flat insoles affected plantar pressures during walking, but did not study kinetic variables at the knee or ankle. Since a shift in the center of pressure beneath the foot has been associated with altered frontal-plane moment magnitudes at the ankle and knee, 10,12 the results presented in the current study where ankle and knee moments were altered with the OSF, SS and SSF control conditions tend to agree with those of McCormick et al

58 The results presented in this study suggest that previous trials utilizing control conditions other than the participant s own footwear should be interpreted with caution. Specifically, it seems possible that some participants in control groups that had an assigned sham intervention may have experienced a biomechanical change when none was intended. As one example, a large randomized controlled trial for knee osteoarthritis was recently conducted where it was determined that lateral wedge insoles, designed to reduce the peak knee adduction moment offered no clinical benefit over flat control insoles. 13 Based on the present study s results for walking, it could be estimated that approximately 47% of the 97 participants evaluated at baseline may have in fact experienced a large change to their knee adduction moments with the control insole, which of course, could have implications on the observed clinical outcomes in the study. Interpretations of the results of the present study are largely based on the assumption that a biomechanical change greater than ±10% is important. This threshold was chosen since (a) intervention insoles/orthotics such as wedged insoles induce changes typically around 5-12%, 10,16,88 and (2) a ±10% change is larger than what is typically reported for within day trial repeatability for joint kinetics variables Therefore, changes from the current study that exceed ±10% can be considered relevant in that they approach or exceed the magnitude of change observed by actual intervention insoles, and that they induce changes greater than what would be expected based on repeatability. These points lend themselves to the question, what is the mechanism by which the tested control conditions alter biomechanical variables? While no conclusive answer can be determined from the existing data, a number of possibilities exist. For instance, the SS condition was a neutral shoe and therefore could have had different supportive/motion control or 43

59 cushioning features compared to the participants own footwear. Since these factors have been shown to affect biomechanical variables during gait, 89,90 it is possible that these factors can explain some of the differences observed across conditions and participants in the present study. Regarding the flat insole, it is possible that cushioning properties were altered, and/or that the added thickness beneath the foot would alter foot landing mechanics, which has been proposed to be related to loading at the ankle and knee during gait. 10,16 These features of the insole, in combination with the different properties of each individual s own shoe, and the possibility of altered proprioception, could have had further interactions that affected gait mechanics as well. Since the primary objective of a control condition is to ensure that the control group in a study is not exposed to a certain intervention or biomechanical modification, it seems to be of paramount importance that precautions be taken in future insole and orthotic studies to ensure that the control condition chosen is biomechanically inert, such that it does not alter participant s normal biomechanics. For clinically based studies, where a change in a biomechanical variable may be associated with clinical outcomes, this is especially critical. If a control condition similar to the OSF, SS or SSF conditions in the present study are chosen for use in future studies, this may necessitate biomechanical evaluation of each participant to ensure they do not experience a large biomechanical change. Depending on the variable of interest, this may result in a very large exclusion rate of participants, therefore requiring a much larger number of individuals to be screened for eligibility. Alternatively, it is proposed that the individual s own footwear be utilized as a control condition. This approach has the advantage that the control groups biomechanics remain unchanged, but is disadvantaged in that it may be difficult to assess placebo effects 44

60 in the experimental group. One possible solution to this is that, if biomechanics are measured for each participant in the experimental group, correlation could be used to assess the relationship between the change in the biomechanical variable of interest, and change in clinical outcome over the study duration for the experimental group. If a significant relationship exists, this would suggest that a change to the biomechanical variable can at least, in part, explain clinical changes that occurred. 23 A limitation of this study is that only healthy, younger adults were evaluated. It remains possible that the proportion of individuals who experience large biomechanical changes for each variable will differ depending on the clinical population of interest (i.e. knee osteoarthritis, patellofemoral pain etc.); however, since healthy individuals tend to show similar biomechanical responses to specialized insole and orthotic designs (e.g. lateral wedge) as clinical populations, 10,91 it is not believed that differences in responses between populations, if any, would be large. Another limitation of this study was that only one standardized shoe, and only one flat insole type was evaluated. It is possible that other designs may yield different results; however, these differences are also not expected to be large since the results in the present study agreed with those from a previous study with slightly different insole preparations. 70 Perhaps more important is the possibility that the biomechanical effects observed at baseline in the present study may change over time, either with shoe/insole wear, or with disease progression or treatment. Additionally, while this study has suggested that certain control conditions may influence important biomechanical variables, considerations such as placebo effects, ascertainment bias (when results/conclusions are distorted by the knowledge of which intervention the participant is receiving), and resentful demoralization (when allocation to a no-treatment group 45

61 influences participant behavior) may also need to be considered in trial designs depending on the primary objective of the trial. 92 Finally, while this study shows support for the idea that participants in other randomized trials for footwear/orthotics may have experienced biomechanical changes when none were intended, it is not possible to conclusively determine if this was in fact the case. Moreover, since the present study found that some individuals experience increases and some experience decreases to biomechanical variables, it is not known if these effects would balance out to a net no effect for a control group. 3.5 Conclusion This study evaluated four different commonly used control conditions for insole and orthotic studies, and found that while consistent directional differences across conditions were not observed, a large proportion of individuals experienced large biomechanical changes relative to what they would normally experience in their own shoe. Based on these findings, it is recommended that the individual s own shoe be utilized for future insole studies as the control condition where biomechanical variables are a primary outcome, and that previous studies using alternate control conditions be interpreted with caution. 46

62 CHAPTER FOUR: PREDICTION OF KNEE JOINT MOMENT CHANGES DURING WALKING IN RESPONSE TO WEDGED INSOLE INTERVENTIONS 4.1 Introduction Knee osteoarthritis (OA) is one of the most common musculoskeletal injuries, and is characterized by elevated frontal-plane knee joint loading (i.e. external knee adduction moments) during walking. 1,6 Indeed, increased frontal-plane knee moments have been associated with increased knee OA pain and greater OA severity. 8,9 Consequently, reducing these moments has become an important strategy for clinical management of knee OA. One of the most common approaches to reducing these frontal plane moments has been to apply wedged footwear insoles or orthotics bilaterally within the patient s shoe. 1,13,88 In biomechanical studies, wedged insoles have been shown to reduce knee adduction moments by 6-15% for most patients; however, for approximately 33% of patients, a negative biomechanical response is observed, i.e. an increase to the knee adduction moment. 16 This fact may contribute to why clinical studies have often shown mixed results in terms of the clinical efficacy of wedged insoles. Given some patients may not respond, biomechanically, to a wedged insole, there has recently been increased interest in being able to identify, prior to insole prescription, which patients are likely to benefit in a biomechanical sense from a wedged insole intervention. 12,16 In research settings, this is being done by only including responders identified through three-dimensional inverse dynamics analysis in study designs, 16 or in testing a variety of insole types to optimize the biomechanical result. 93 However, in clinical settings, where expensive gait analysis equipment is typically unavailable, there is 47

63 currently no way to predict if a patient is likely to experience a reduced KAM in response to a wedged insole. Research studies have found that center of pressure positions beneath the foot, 12,20 knee joint lever arms, 12 or ankle angles at touchdown, 16 may be weak-moderate predictors of biomechanical efficacy; however, these methods all require a detailed and expensive gait analysis setup. For a prediction method to be clinically relevant and useful in community settings, the method must be small in design so as not to take up clinic space, have the potential to collect relevant data with inexpensive equipment such that it is affordable by specialty clinics, support relatively rapid data collection and analysis time, and be a strong predictor of the expected biomechanical response to a wedged insole. Therefore, the purpose of this study was to develop and show proof of concept for a method for predicting the expected change to the frontal-plane knee joint moment during walking with a wedged insole. Based on Newton-Euler equations of motion, 50 it can be seen that the variables that contribute to the magnitude of the knee joint moment include inertial parameters of the leg, angular velocities and accelerations of the leg, forces acting at the ankle and knee, moments acting at the ankle, and lever arms from the ankle and knee joints to the segment centers of mass. We hypothesized that by focusing exclusively on the mediolateral components of the lever arms of the leg at a single time point during a movement similar to walking, the expected frontal-plane knee moment change could be predicted. 48

64 4.2 Methods Participants Fifteen healthy individuals without any history of musculoskeletal injury (10 males, 5 females, mean±sd age of 24.9±4.5 years, height of 174.7±10.1 cm, mass of 72.1±14.0 kg), and 19 individuals with medial knee osteoarthritis (5 males, 14 females, mean±sd age of 59.8±6.7 years, height of 170.5±10.7 cm, mass of 89.7±23.6 kg), as diagnosed by a physician according to the American College of Rheumatology radiographic and clinical criteria, 94 participated in the study Data Collection The protocol was approved by the University of Calgary s Conjoint Health Research Ethics Board, and all participants gave informed written consent prior to any data collection. Three retroreflective tracking markers were secured to each of the foot (shoe) and leg segments. For healthy individuals, this was always done on the right lower extremity. For individuals with knee OA, this was always done on the most symptomatic lower extremity. In a randomly assigned order, participants completed 5 trials walking along a 20m runway in the control condition (participant s own shoes), in a medial wedge condition, and in a lateral wedge condition, in both cases where a 6mm wedge was applied bilaterally beneath the sock liner of the participant s own shoes. 93,95 A force platform (Kistler Group, Winterthur Switzerland) mounted flush with the lab floor collected ground reaction force data in three dimensions at a frequency of 2400 Hz and an eight camera Motion Analysis 49

65 system (Motion Analysis Corp., Santa Rosa, CA) collected 3D retroreflective marker trajectories at a frequency of 240 Hz during each trial. Photocells placed 1.9 m apart were used to monitor gait speed for each trial, where healthy participants maintained speeds of 1.5 m/s (±5%) and OA individuals maintained speeds of 1.3 m/s (±5%) for each trial. Different gait speeds were chosen as individuals with knee OA tend to walk slower than uninjured individuals. 17,60 If the speed requirements were not met, or if the participant did not land on the center of the force platform with the lower extremity of interest, the trial was repeated until a total of five successful trials were obtained. Additionally, each participant completed five single step trials with each of the three footwear conditions. These trials were not speed controlled, and consisted of the participant standing in a neutral position with both feet just in front of the force plate, and then taking a single step over the force plate (Figure 4.1). This included landing and pushing off on the plate with their lower extremity of interest and then returning to a neutral standing position on the other side of the force plate. Finally, additional retroreflective markers were placed over the medial and lateral malleoli and epicondyles of each participant, and a standing neutral trial was collected, where the participant stood on the force platform in the anatomical position. This was done for each of the footwear conditions for each participant. 50

66 Figure 4.1. The single step procedure is shown. (1) The participant aligns themselves in a neutral position in front of the force plate. (2) The participant takes a step at a self-selected speed over the force plate, ensuring to land on the force plate with the limb of interest. (3) The participant steps off the plate and re-aligns to neutral stance on the opposite side of the force plate. In (2), when the vertical ground reaction force is at its maximum in the first half of stance phase, the mediolateral positions of the knee joint center, leg center of mass, ankle joint center, and foot center of mass are extracted. In the figure above, the sagittal plane is shown. 51

67 4.2.3 Data Processing Kinematic and kinetic data for the walking, single step and neutral trials were imported into KinTrak (v7.0, University of Calgary, Calgary, AB), and smoothed using fourth order Butterworth low-pass filters with cutoff frequencies of 12 Hz and 50 Hz, respectively. 46,95 From the neutral trials, the knee and ankle joint centers were defined as the point 50% of the distance between the epicondyle and malleoli markers, respectively. 51 These locations were used to define segment lengths for the foot and leg. Segment center of mass locations, segment masses and segment moments of inertia were taken from the literature, or calculated from the neutral trial using proportions defined in the literature. 53,54 This was done for each neutral trial collected, and these lower extremity models were then applied to their associated walking and single step trials (e.g. lateral wedge neutral trial applied to walking and single step lateral wedge trials for each participant). For walking and single step trials, stance phase was defined as the period from foot touchdown to foot takeoff, which were defined as the rising cross of 9.81 N and the falling cross of 9.81 N, both of the vertical ground reaction force. In walking trials, external knee adduction moments were calculated during stance phase using a standard Newton-Euler inverse dynamics approach, 50 and the peak knee adduction moment that occurred during the first 50% of stance phase was extracted for each trial. For each participant, the mean peak knee adduction moment was calculated for each footwear condition. 52

68 For single step trials, only the mediolateral kinematic data and vertical ground reaction forces were studied. First, the time at which the vertical ground reaction force was at its maximum within the first 50% of stance phase was identified. This time point was chosen because it roughly coincides with the time at which the first peak knee adduction moment occurs, and is an easily identifiable marker on the vertical ground reaction force curve. At this time point, the mediolateral positions of the knee joint center, leg center of mass, ankle joint center and foot center of mass were extracted for each trial. For each participant, the mean mediolateral positions of each marker were then calculated for each footwear condition Statistical Analysis All statistical analyses were performed in MATLAB r2015a (MathWorks Inc., Natick, MA) at a significance level of From the mean data for each participant, the percent change in walking knee adduction moment, and percent change in single step marker positions were determined for each footwear condition. These percent changes were expressed relative to the neutral condition. Four two-tailed multiple linear regression equations were then developed; two for the healthy study group, and two for the knee OA study group. The first multiple regression (MarkersMW) studied changes induced by the medial wedge insole, where percent changes to the single step mediolateral knee joint center, leg center of mass, ankle joint center, and foot center of mass positions were included as independent variables, and percent change in peak knee adduction moment during walking induced by the medial wedge was the dependent variable. The second multiple regression (MarkersLW) studied changes induced 53

69 by the lateral wedge insole, where percent changes to the single step mediolateral knee joint center, leg center of mass, ankle joint center, and foot center of mass positions were included as independent variables, and percent change in peak knee adduction moment during walking induced by the lateral wedge was the dependent variable. As mentioned previously, this procedure was performed separately for both the healthy group, and the knee OA group such that knee OA single-step kinematics were assessed against knee OA walking moments, and healthy single-step kinematics were assessed against healthy walking moments. This allowed us to determine if similar relationships existed for both healthy and knee OA individuals, and so the two study groups were never directly compared to each other. Given the low sample sizes used to develop the regression equations, we conducted a simulation test whereby the stability of these regression models, or the ability to predict future data, was evaluated. This was done by computing the PRESS statistic, and corresponding Predicted R-Squared (r 2 p) value an indication of how the model will hold to predictions with new, future data 96 as follows: first, the MarkerMW and MarkerLW equations are redeveloped n times for each study group, where n is the sample size of each study group (n=15 for healthy and n=19 for knee OA), each time with n-1 observations. Next, the single-step raw data from the removed subject are applied to the new equation to output their predicted change in knee adduction moment during walking. This process is then repeated for all subsequent subjects. For example, in the first iteration of MarkerLW for the healthy group, subject 1 is removed, the regression equation is developed using subjects 2-15, and then single-step data from subject 1 is input into the model where predicted change in KAM with the lateral wedge is output. Then, subject 2 is removed, and 54

70 the MarkerLW equation is developed using subjects 1, 3-15, and subject 2 s data is input into the model etc. This procedure is done for all subjects, in both the knee OA and healthy groups, and for both insole types. The result from this procedure is a predicted KAM change during walking for all participants for both insole types. By assessing these predicted values against the actual walking KAM changes, the Predicted R-Squared (r 2 p) is determined using two-tailed linear regression. This analysis does not provide indication of how strong the predictor variables from the single step procedure are at predicting change in KAM (this is determined from the standard regression approach described in the previous paragraph). Instead, this procedure highlights whether the model, in its current form, can be used to predict new data. Based on the predicted knee adduction moment changes output by the simulation/predicted R-Squared procedure, an algorithm was developed that would assess the predicted changes with lateral and medial wedges for all participants and either indicate lateral wedge, medial wedge, or no wedge as a recommendation for each participant (Figure 4.2). In cases where a lateral wedge or medial wedge recommendation was made, this recommendation was classified as correct if the recommended insole did indeed reduce the knee adduction moment during the walking trials, and incorrect if the recommended insole increased knee adduction moments. In cases where no insole was recommended, this was classified as a correct recommendation if indeed neither insole type reduced knee adduction moments during walking, and classified as incorrect if in fact one of the insole types would have reduced knee adduction moments. 55

71 Figure 4.2. Flow chart showing the data collection and analysis algorithm. 56

72 4.3 Results For the healthy individuals, a significant relationship was found for MarkerMW (R 2 =0.67, p=0.016), and also for MarkerLW (R 2 =0.72, p=0.008) for predicting the change in knee adduction moment induced by a wedge using mediolateral lower extremity marker positions. For the knee OA individuals, a significant relationship was found for MarkerMW (R 2 =0.54, p=0.020), and also for MarkerLW (R 2 =0.52, p=0.026) for predicting the change in knee adduction moment induced by a wedge using mediolateral lower extremity marker positions. When comparing predicted knee adduction moment changes against actual knee adduction moment changes during walking, a significant relationship was found in the healthy study group for lateral wedge induced changes using MarkerLW (r 2 p=0.44, p=0.007), and a near significant relationship was found for medial wedge induced changes using MarkerMW (r 2 p=0.21, p=0.084). A significant relationship was found in the knee OA group for lateral wedge induced changes using MarkerLW (r 2 p=0.30, p=0.016), but no significant relationship was found for medial wedge induced changes using MarkerMW (r 2 p=0.10, p=0.19). These results are shown in Figure 4.3, and details regarding the equations of the lines of best fit for the predicted vs. actual KAM changes are shown in Table 4.1. Also provided in Figure 4.3 are the 95% confidence intervals, which show how the determined line of best fit in the present study may compare with the true line of best fit for this method, and also the 95% prediction intervals, which show where subsequent data points can be expected based on the current prediction model. 57

73 Figure 4.3. Relationship between predicted knee adduction moment (KAM) changes generated from the simulation analysis and actual changes to the knee adduction moment as measured during walking. The predicted r-squared value (r 2 p) is shown with lines of best fit. Details on the lines of best fit are shown in Table 4.1. The dotted orange lines show the 95% confidence interval of the line of best fit, and the dotted blue lines show the 95% prediction interval. 58

74 Table 4.1. Slopes (ß1) and intercepts, with 95% confidence intervals shown in brackets, are shown for each of the lines of best fit produced for the simulated predicted KAM changes during walking vs actual KAM changes during walking, that are shown in Figure 4.3. Walking KAM Prediction Simulation ß1 (95% C.I.) Intercept (95% C.I.) r 2, p-value Healthy Lateral 0.6 (0.2 to 1.0) -3.8 (-11.4 to 3.9) r 2 =0.44, p=0.007 Healthy Medial 0.4 (-0.1 to (-7.64 to 13.6) r 2 =0.21, p=0.084 OA Lateral 0.4 (0.1 to 0.8) -4.2 (-7.6 to -0.8) r 2 =0.30, p=0.016 OA Medial 0.3 (-0.2 to 0.8) 0.2 (-3.8 to 4.2) r 2 =0.10, p=

75 The prediction error rate (situations where the predicted knee moment change was opposite in direction to the actual knee moment change during walking) for the MarkerMW equations were 2/15 (13%) for the healthy individuals and 9/19 (47%) for the knee OA individuals. The prediction error rate for the MarkerLW equations were 6/15 (40%) for the healthy individuals and 2/19 (11%) for the knee OA individuals. When utilizing predicted outputs for both the MarkerLW and MarkerMW equations to identify the recommended wedged insole condition for each participant, we could correctly identify, for 12/15 healthy individuals and 17/19 knee OA individuals, the correct recommendation (either insole or no insole) to ensure knee adduction moments were either reduced, or prevented from increasing. In the cases where the algorithm made incorrect recommendations in the healthy group, the algorithm suggested no insole on one occasion when in fact a lateral wedge would have reduced the knee adduction moment, and recommended lateral wedge on two occasions when this would have in fact resulted in an increased knee adduction moment. In the knee OA group, there was one case where no insole was recommended when in fact a medial wedge would have been the best recommendation, and one case where a medial wedge was recommended when in fact no insole would have been the correct recommendation. 4.4 Discussion The purpose of this study was to develop a new approach for predicting the biomechanical effects of wedged insoles during walking. It was hypothesized that mediolateral positions of the leg at the instant of the maximum vertical ground reaction force during the first half of stance during a single step movement would be related to 60

76 change in knee adduction moment induced by a wedged insole intervention. Our primary analyses confirmed this hypothesis, where significant relationships were found for predicting the effects of medial and lateral wedge insoles for both healthy and knee osteoarthritic individuals. Previous research has suggested relationships between change in knee adduction moment during walking and change in center of pressure position, change in frontal-plane ankle angle or change in knee to center of pressure lever arm magnitudes; 12,16 however, these associations were typically quite low, and were developed using walking multi-axis data collection methods during controlled walking movements. Thus, the large R 2 values presented for the four regression equations that utilized uniaxial data, during a simpler and less controlled movement are noteworthy. This is of critical importance for the implementation of these findings into a tool that can be utilized in clinical settings. Additionally, given similar results were found for a young healthy population (mean BMI of ~23.6 kg/m 2 ) and for an older, knee osteoarthritic population (mean BMI of ~30.8 kg/m 2 ), the method seems applicable to a broad range of individuals. The use of uniaxial force measurement in our method significantly lowers cost relative to multi-axis systems that are required for computation of the center of pressure beneath the foot. Since the force variable chosen is a very distinguishable peak in the vertical ground reaction force, identification of this peak and associated time point using force plates that do not have a high sampling frequency will also be possible. Since our method uses only 2D, frontal-plane positions of the leg, and during a fairly slow movement (i.e. single step), a relatively simple data collection setup such as a standard video camera 61

77 or X-Box Kinect system could likely be utilized for collection of this data at a low cost; however, the validity of using these systems needs to be further confirmed. One conceptualization of this system would be that a participant wears an identification band around their knee and ankle to signify joint locations, from which the segment center of mass positions are computed using literature data. Then the individual takes a series of single steps while a video system collects frontal plane motion data and a uniaxial force platform synchronously collects vertical forces (not unlike a digital scale with a real-time feed). At the instant of maximal vertical force, which could be easily detected by a simple computer algorithm, the limb positions are obtained. Once repeated with a different footwear type (e.g. a wedged insole), the difference between the two conditions is automatically calculated and input into standardized regression equations to yield the predicted knee adduction moment outcome. While significant relationships were found for all four regression equations, only the lateral wedge equations remained significant when evaluating the predicted r 2 p values. The likely reason as to why only lateral wedge predicted r 2 p values were found to be statistically significant in the present study is that the natural variability of lateral wedges is lower than medial wedges. For instance, typically 70% of individuals experience a KAM reduction with a lateral wedge, 16 whereas only 40-50% of individuals experience a KAM reduction with medial wedges. 66 However, since our initial regression equations using step data to predict walking KAM changes were all statistically significant, this indicates that with larger sample sizes the equations may be refined such that the variability in using the equations to predict medial wedge outcomes may be reduced. Given the r 2 p values are generally lower, larger sample sizes would be beneficial for the lateral wedge equations as 62

78 well. In general, larger sample sizes will also improve the confidence and prediction intervals for this method, allowing for improved estimation of the method s potential efficacy. Currently, the confidence interval of the line of best fit is of greater importance than the prediction interval, as the model is in proof of concept stage, and not yet intended for prediction of new data. While the prediction algorithm benefits from utilizing both the medial wedge and lateral wedge equations to correct for possible error with one insole type, increasing the predictive capacity of both equations would likely reduce the likelihood of total error in assigning wedges. Based on the current data for the lateral wedge equations, error rates (11-13%) were much lower than what is typically reported in the literature when all subjects are given a lateral wedge (33%), 16 suggesting the regression equations offer an advantage over current clinical and research practices of assigning all individuals a lateral wedge. In addition, given the demographics of the two groups evaluated were quite different, the finding that the method could be used in both groups is promising. Certainly, these simulated predictions offer only a proof of concept analysis, and refinements with larger sample sizes should be made to build a more robust regression model. When using the results of two equations to predict the type of insole recommended, the analysis performed fairly well. Some of the error in the results reported here stems from the reduced predictive capacity of the medial wedge conditions, and so it is expected to improve as refinements are made with larger sample sizes. With larger, more robust regression equations, more detailed analyses of true-positive, true-negative, false-positive and false-negative rates can be determined. 63

79 A source of error in this study was that marker positions at the knee and ankle were reapplied with each new footwear condition rather than remaining fixed to the participant throughout all conditions and trials. While markers were placed on each participant by the same researcher in all occasions, slight changes in marker positions are still possible, which would introduce error to the joint center locations. This error could be mitigated in the future by ensuring markers remain fixed to the participant during all trials and footwear conditions. Additionally, while only uniaxial force and 2D position data were utilized in the current study, these data were collected using sophisticated 3D systems meant for full gait analysis. Thus, future work will need to establish how this method of prediction functions with simpler, less expensive equipment. Finally, previous research has shown good agreement in terms of knee moment magnitudes between ipsilateral and contralateral knees during gait in individuals with knee OA, 97 and thus we assume the results of the present study would be applicable regardless of which knee is tested. However, since we only evaluated the most symptomatic knee for individuals with OA, and only the right knee for healthy individuals, we cannot conclusively determine if this is actually the case. 4.5 Conclusion It has been shown that changes to the knee adduction moment during walking resulting from a wedged insole intervention can be predicted using mediolateral position data of the lower extremity during a single step movement. While still in early stages of development, future refinements of this method may allow for simple biomechanical data collection and analysis in clinical settings, allowing for a personalized approach to wedged insole prescription. 64

80 CHAPTER FIVE: REDUCED KNEE ADDUCTION MOMENTS USING WEDGED INSOLES FOR MANAGEMENT OF MEDIAL KNEE OSTEOARTHRITIS: A 3-MONTH RANDOMIZED CONTROLLED TRIAL 5.1 Introduction Osteoarthritis (OA) is a degenerative disease of the cartilage that commonly affects the medial tibiofemoral compartment of the knee. 4 As no cure exists for knee OA, and knee joint replacement is typically reserved as a last line option, clinical management is often aimed at conservative non-surgical strategies. 5,6 Biomechanically, reduction of medial compartment load, often quantified as the knee adduction moment, 6,30 has been a primary goal of conservative management given that increased KAMs have been related to OA severity, pain, and disease progression. 8,9 Despite these associations between KAM magnitude and disease severity, evidence regarding the association between KAM reductions and improved clinical outcomes is lacking. Laterally wedged footwear insoles/orthotics have been shown to, on average reduce KAMs, 16,17,65 and have therefore been the subject of much research for management of knee OA over the past 15 years. Traditionally, a lateral wedge would be applied to all study participants without actually measuring their KAMs, 13,22 and it would simply be assumed that KAMs were reduced for all participants; however, recent evidence has highlighted than in fact up to 33% of individuals receiving a lateral wedge actually experience increased KAMs during walking with lateral wedges. 16 This finding may partly explain why the clinical effectiveness of lateral wedges has been mixed across trials, and highlights the notion that the influence of reduced KAMs on clinical outcomes is not presently known, 65

81 because it has not specifically been controlled for. In the short term, one study failed to show an association between KAM reduction and immediate, same day pain reduction; 68 however, it is not known if this would hold true for longer follow-up durations. Like lateral wedges, medial wedges also affect KAMs during gait, 10,64,66 and may induce an increased KAM or decreased KAM depending on the individual. Consequently, this wedge type may also serve as an appropriate intervention for patients who experience KAM reductions with this wedge type. Considering the lack of evidence regarding the effects of reduced KAMs on clinical outcomes for individuals with medial knee osteoarthritis, a 3-month randomized controlled trial was undertaken, comparing wedged insoles (KAM reduction) to control footwear (no KAM change) to test the hypothesis that pain reduction over three months would be associated with individuals experiencing reduced KAMs. Moreover, it was hypothesized that pain reduction would have a dose-response relationship to KAMs, where larger KAM reductions would be associated with larger pain reductions. 5.2 Methods Study Design The detailed methodology of this study has been described previously (Appendix A); 93 however, any modifications to the previously described protocol have been outlined in this chapter. The full trial flow can be seen in Figure

82 Figure 5.1. Participant flow through the study. 67

83 The study was conducted at the University of Calgary between January 2015 and October The study was approved by the Conjoint Health Research Ethics Board of the University of Calgary. Written informed consent was obtained from all participants prior to any data collection, testing, or evaluation of medical histories. The study was registered with ClinicalTrials.gov (ID# NCT ). The sample consisted of individuals who were recruited from the greater Calgary, Alberta area by poster, television and social media advertisement between January 2015 and July Inclusion criteria were age between 40 and 85 years, Knee Injury and Osteoarthritis Outcome Score (KOOS) of 75 or lower on the pain subscale on initial contact, confirmed diagnosis of unilateral or bilateral knee OA based on the American College of Rheumatology criteria, 94 and confirmation that medial compartment disease was the primary location of symptoms based on clinical exam by a physician. Patients must also have had an x-ray taken of their most symptomatic knee from within the two years prior to the clinical assessment. These x-rays were originally examined by a sport medicine physician for confirmation of radiographic disease. Exclusion criteria were existing knee replacement, knee surgery within past six months, musculoskeletal or neuromuscular injury to knee within past two months, as well as any of the following management strategies assessed at baseline: regular use of walking aids in past two months, unloader knee braces within the past two months, corticosteroid injection to the knee within the past three months, or other viscosupplement injection within the past six months. All individuals meeting these criteria had their x-rays of their most symptomatic knee further evaluated by a radiologist, who assigned a Kellgren-Lawrence grade. 98 Contrary to our original proposal, 46 individuals of any KL grade 1 were included in the study. This 68

84 allowed for inclusion of all individuals of any severity with OA primarily localized to the medial compartment on clinical exam, and who were deemed inappropriate for surgery, yet symptomatic at the time of the study. X-rays were also quantified on the basis of unilateral or bilateral disease, as well as uni-compartmental, bi-compartmental, or tricompartmental disease Footwear Usual Footwear Usual footwear was defined as the footwear that the participant had used most regularly over the past two months. Although use of an orthotic/insole was described as an exclusion criteria previously, 93 this approach was modified to consider previous orthotic/insole use as acceptable under the assumption that if this is what the individual was used to wearing, a reduction in KAMs relative to this usual footwear would still theoretically be beneficial Wedged Insoles The wedged insoles evaluated were 6 mm laterally wedged insoles, and 6 mm medially wedged insoles. In both cases, insoles ran the length of the foot, 18 and were fabricated using a 3D printer (New Balance Athletic Shoe Inc., Boston, MA). When evaluating insoles at baseline, and when utilizing insoles during the three month study period, insoles were applied bilaterally in the participant s usual footwear. If this footwear had a sock liner, insoles were placed beneath the sock liner. If this usual footwear included 69

85 an orthotic/insole that the individual was already using, this orthotic/insole was removed when the experimental insoles were applied Procedures & Measurements Clinical Assessment Initial screening for location of pain, exclusion criteria and baseline KOOS pain score was carried out by an investigator by telephone and . For participants meeting these initial criteria, x-rays from within the past two years were then reviewed. Pending eligibility requirements were still met, participants were invited for a physical assessment by the sport medicine physician where eligibility was confirmed. The time from initial contact to scheduling a physician consultation was typically between 2-6 weeks Baseline Testing All participants meeting clinical eligibility underwent baseline testing in a biomechanics laboratory, usually between 1 and 14 days following the physician consultation. The KOOS full-version, 99 physical activity scale for the elderly (PASE), 100 and UCLA physical activity scale surveys were administered. 101 Additionally, a report of co-intervention use, such as NSAIDs, tensor sleeves, targeted knee exercise etc., over the past week was provided. The KOOS survey included subscales of pain, symptoms, activities of daily living function (ADL), sport and recreation function (Sport/Rec), and quality of life (QOL). For the KOOS survey, if the participant experienced bilateral OA, they were asked to provide scores relative to their most symptomatic knee. A history survey was also administered that recorded participant history of surgery to their most 70

86 symptomatic knee. Finally, participant baseline metrics such as age, height and body mass were recorded. The full versions of the baseline questionnaires used in the study are available in Appendix A and Appendix C. Each participant also underwent dual x-ray absorbtiometry (DXA) testing, where total body fat percentage, and whole body bone density were determined. Biomechanical gait analysis during walking was performed on all participants while wearing their usual footwear, usual footwear with the lateral wedge insole, and usual footwear with the medial wedge insole. The procedures are described in full in Appendix B. 93 Briefly, participants walked along a runway in a motion analysis lab with each footwear type at a controlled speed of 1.3±0.07 m/s while 3D ground reaction forces and 3D lower extremity trajectories were recorded. Following the tests with each footwear condition, participants rated their perceived overall comfort using a 100 mm visual analogue scale. A neutral trial was collected for each footwear condition, and ankle, knee and hip joint center locations, foot and leg segment masses, centers of mass, and moments of inertia were defined as described elsewhere. 46,51 The first peak knee adduction moment (KAM) during stance phase was the primary outcome measure of knee joint load, and was quantified immediately following the completion of biomechanical testing for each participant s most symptomatic knee using an inverse dynamics approach to determine the effects of the wedged insoles. 50,51 Participants who were found to experience KAM increases with both the lateral wedge and medial wedge were then eliminated from further study. Thus the study consisted of only biomechanical responders to wedged insoles. Additional biomechanical variables were also calculated. These included the knee adduction angular impulse (KAAI) during stance phase, the 3D resultant moment at the 71

87 same point in time as the KAM (square root of the sum of squares of the three moment components), the amount of varus thrust during stance phase (i.e. maximum mediolateral displacement of knee during stance relative to touchdown position), as well as the frontalplane ankle joint angle (measured from floor to ankle to knee), frontal-plane knee joint angle (measured from ankle to knee to hip), and left to right frontal-plane knee joint spacing (i.e. distance from left knee to right knee) during neutral stance. The latter three measures of neutral alignment were performed while the participant stood in the anatomical position with feet positioned 18 cm apart (Figure 5.2). Each of these variables were quantified for each footwear condition and participant Randomization, Blinding and Allocation Following baseline data collection, participants were randomized using blockrandomization sequences in a 1:1 ratio that were generated using a computer program, and stratified based on sex, to either a wedged insole group, or a waitlist control group. Investigators nor participants were aware of which group the participant was randomized during baseline data collection, as randomization occurred at the end of the laboratory testing session. At this time, both the participant and investigator were aware of the randomization. Once participants were randomized, those allocated to the insole group were given either a medial or lateral wedge (whichever reduced KAMs the most). While medial wedges have not been utilized previously for knee OA, if they reduce KAMs during walking then theoretically they should be beneficial. Participants in the waitlist control group were informed that they would be receiving either the medial or lateral wedge (whichever reduced KAMs the most) in three months. Investigators involved in statistical 72

88 analysis were blinded to which group the participant was allocated by providing numeric data rather than group condition text Follow-Up Following randomization, participants were asked to use their assigned intervention, as tested, as much as possible over the next three months, and to avoid transferring the insole (if in wedged insole group) to other footwear during that time. Both groups were asked to avoid the use of any new co-interventions. The KOOS, PASE, UCLA and co-intervention survey were completed by the participant and returned to the investigators by at three months follow-up. Additionally, at this time point, participants responded to questions regarding their frequency of use of their assigned footwear over the past week, as well as whether new injuries had developed. The full version of the follow-up survey is found in Appendix C. 73

89 Figure 5.2. A frontal-plane diagram is shown for each of the static alignment variables evaluated. Details on the method of obtaining joint center positions are available elsewhere. 46,51 74

90 5.2.4 Sample Size Calculations Sample size calculations were performed based on the primary outcome variable, the KOOS pain score, with a desired power of 80% and a significance level of 0.05 using the statistical software Stata version 13 (Stata Corp., College Station, TX). Given the primary objective was to identify the dose-response relationship between KAM reduction and change in pain, sample size calculations were based on within group comparisons for the wedged insole group, and between group analyses were secondary. The minimum perceptible clinical improvement and minimum detectible change in KOOS score has been reported to be about 13.4 points. 99,102 Using a baseline KOOS pain score of 56.5 points, 103 and a standard deviation of 15 points, 102,103 to detect a minimal clinically important improvement of 13.4 points within each group, 20 participants are required in each group, or 40 total participants. To account for a potential dropout/exclusion rate of 15%, 13 recruitment of a minimum of 46 participants was planned Statistical Analysis All data were analyzed on an intention-to-treat basis using Stata version Through visual inspection of boxplots of each variable, all data were determined to be normally distributed, and thus parametric tests were used for all data analysis. For all tests, a significance level of 0.05 was used. At baseline, participant characteristics that were continuous variables (eg. age, height, mass, BMI, DXA body fat percentage, bone density, and mass in 20 s) were 75

91 compared across the insole, control and excluded participants using ANOVA. All other categorical baseline variables were compared across the insole, control and excluded participants using Chi-Square. Biomechanical data and clinical outcomes data were compared between the insole and control groups at baseline using MANOVA with twotailed independent samples t-tests for post-hoc assessment. Within the wedged insole group, biomechanical effects of the intervention insole were determined by calculating mean differences between usual and intervention footwear and 95% confidence intervals. Differences in KOOS pain, the primary outcomes variable, over the three month period between footwear groups were assessed using ANCOVA, adjusting for baseline KOOS pain score. 105 To account for missing follow-up data (n=5/38 participants), a multiple imputation was utilized to fill missing data, and a sensitivity analysis was performed by comparing ANCOVA results with imputation to ANCOVA results obtained from exclusively completed cases. 104 As results were unchanged, results are presented for completed cases only. Assessments between groups for all follow-up variables including KOOS, PASE and UCLA scores are expressed as mean differences with 95% confidence intervals. To assess the effects of the intervention on KOOS pain exclusively for the wedged insole group, a two-tailed paired samples t-test was used. To assess whether a difference existed between groups in terms of the number of participants experiencing a clinically relevant improvement in KOOS pain of 13.4 points, a Chi-Square was used. Multivariable linear regressions were conducted on the wedged insole group to evaluate whether changes in KAM, KAAI, 3D resultant moment induced by the intervention footwear were associated with change in KOOS pain over three months. 76

92 Covariates assessed in these models included adiposity by DXA scan, Kellgren-Lawrence grade, baseline KOOS pain and change in PASE score over three months. 5.3 Results Demographics Of the 367 volunteers interested in the study, 49 (13.3%) met all eligibility criteria and agreed to participate (Figure 5.1). In total, 48 individuals completed baseline testing, while one participant did not complete the necessary testing. Of the 48 participants tested at baseline, 10 (20.4%) did not experience KAM reductions with either insole type and so were excluded on the basis of biomechanical ineligibility, resulting in 19 individuals being randomized to the insole group and 19 to the control group. Participant baseline characteristics are shown in Table 5.1, which includes characteristics of the 10 participants who were biomechanically ineligible for further participation. As can be seen, the groups were generally well matched. 77

93 Table 5.1 Baseline characteristics for the two study groups, as well as the group of participants excluded following biomechanical testing are shown as number (%) unless denoted by *, where values are means (SD). P-values are shown for across group comparisons from Chi-square (X 2 ) for proportion data, and one-way ANOVA for continuous data. Variable Wedged Insole (n=19) Control (n=19) Excluded (n=10) P-value Female 13 (68) 11 (58) 7 (70) *Age, years 59.9 (7.4) 59.6 (7.7) 59.7 (9.9) *Height, m 1.69 (0.12) 1.73 (0.09) 1.64 (0.10) *Body mass, kg 93.3 (23.8) 87.6 (20.8) 77.6 (20.0) *BMI, kg/m (8.0) 29.2 (6.7) 28.5 (5.9) *Body fat, % 37.4 (8.5) 33.0 (8.9) 35.3 (8.8) *Bone density, g/cm (0.14) 1.10 (0.12) 1.05 (0.15) Bilateral OA 18 (95) 16 (84) 7 (70) Compartments Uni-compartmental OA 2 (11) 2 (11) 1 (10) Bi-compartmental OA 3 (16) 2 (11) 1 (10) Tri-compartmental OA 14 (74) 15 (79) 8 (80) Radiographic Severity K-L Grade 1 5 (26) 4 (21) 1 (10) K-L Grade 2 2 (11) 4 (21) 2 (20) K-L Grade 3 3 (16) 2 (11) 1 (10) K-L Grade 4 9 (47) 9 (47) 6 (60) Duration of symptoms to <5 years 9 (47) 10 (53) 2 (20) 5 to <10 years 5 (26) 6 (32) 2 (20) 10 years 5 (26) 3 (16) 6 (60) Hx of knee surgery 2 (11) 5 (26) 5 (50)

94 5.3.2 Footwear & Biomechanics Of the initial 48 participants who underwent baseline testing, 37 (77.1%) experienced KAM reductions with the lateral wedge, and 20 (41.7%) experienced KAM reductions with the medial wedge. In the wedged insole group, lateral wedges were found to reduce KAMs the most for each of these participants, and so the wedged insole group consists entirely of lateral wedges. Of these 19 individuals in the wedged insole group, who by experimental design all had KAM reductions, 15 (78.9%) experienced reduced KAAIs, 13 (68.4%) experienced reduced 3D resultant moments, 13 (68.4%) experienced reduced varus thrust, 14 (73.7%) experienced a more valgus knee joint angle during neutral stance, 15 (78.9%) experienced a more everted ankle angle during neutral stance, and 11 (57.9%) experienced a reduced distance between their right and left knee joint center during neutral stance. In the wedge insole group, there were 8 (42.1%) participants who experienced increased comfort with their assigned insole. Differences between baseline usual footwear biomechanics between the wedged insole and control groups were not found to be statistically significant (F(8,29)=0.87, p=0.56). A summary of all biomechanical and footwear data can be found in Table

95 Table 5.2. Biomechanics and footwear comfort data are shown as means (SD) between groups for the usual footwear condition, and also within the wedged insole group, where usual footwear is compared to the intervention insole. Mean differences (95% confidence interval) are shown for the wedged insole group, where negative values indicate a lower wedged insole result vs usual footwear. Variable KAM, Nm Usual Footwear 68.1 (21.6) Wedged Insole (n=19) Wedged Mean Difference Insole (95% CI) (-10.6 to -5.1) (18.6) 1 Control (n=19) Usual Footwear 59.2 (18.2) KAAI, Nms 30.5 (10.7) 28.1 (10.0) -2.4 (-3.6 to -1.1) (8.5) 3D Resultant, Nm 72.4 (19.9) 66.4 (17.7) -6.0 (-9.5 to -2.6) (18.9) Varus Thrust, mm 41.5 (16.9) 38.0 (17.4) -3.5 (-9.2 to 2.3) 36.3 (20.9) Static Ankle Angle, deg (15.5) (15.9) 7.9 (-0.1 to 16.0) (15.1) Static Knee Angle, deg (4.7) (4.8) 1.1 (-0.6 to 2.8) (4.6) Static Left to Right Knee Joint Spacing, mm (134.3) (137.5) 0.3 (-3.9 to 4.6) (128.8) Footwear comfort, mm 66.1 (21.2) 57.5 (32.6) -8.6 (-25.2 to 8.1) 66.7 (24.9) *No significant differences were found between control and wedged insole groups for usual footwear. 1 p<0.001, 2 p=0.002, 3 p=

96 5.3.3 Follow-up Outcomes All follow-up clinical data can be found in Table 5.3. When assessing baseline KOOS, PASE and UCLA scores between groups by MANOVA, a significant difference was detected (F(7,30)=2.52, p=0.036); however, post-hoc assessments were not found to be statistically significant (KOOS ADL was bordering on significance at p=0.050). Change in KOOS pain over 3-months was not significantly different between the insole and control group when adjusted by baseline pain (F(1)=0.49, p=0.49). The model was unchanged when adjusting by sex, BMI, DXA Body fat, Kellgren-Lawrence grade, PASE change over three months, or total assigned intervention usage at three months (calculated as days per week multiplied by hours per day). When assessing change in KOOS pain exclusively for the wedged insole group, there was no statistical difference between scores at baseline and three months (p=0.173). In the wedged insole group, 5/19 individuals experienced a clinically meaningful improvement in KOOS pain, and in the control group 2/19 individuals experienced a clinically meaningful improvement in KOOS pain; this was not a significant difference (X 2 (1)=1.58, p=0.209). When assessing variables related to KOOS pain change over three months for the wedged insole group (n=15), the best model was found to include baseline KOOS pain and change in PASE score over three months as independent variables (R 2 =0.57, p=0.007), with the majority of the variance explained by baseline pain. Specifically, the relationship was such that improved KOOS pain over three months was associated with worse baseline pain (i.e. a lower KOOS score) and reduction in physical activity (i.e. a lower PASE score) over three months. When testing this model for the control group, no association was found (R 2 =0.04, p=0.737). Interestingly, no relationships were found between KOOS change in 81

97 pain and KAM reduction (R 2 =0.02, p=0.595), KAAI reduction (R 2 =0.03, p=0.844), or 3D resultant moment reduction (R 2 =0.004, p=0.814). Scatter plots showing the loading vs. change in pain analyses are displayed in Figure

98 Figure 5.3. Relationship between change in KOOS pain over 3 months and (A) change in KAM, (B) change in KAAI, and (C) change in 3D resultant moment are shown. 83

99 Table 5.3. Clinical outcomes data at baseline and 3 months follow-up. Mean differences within groups are based on complete cases only (n=15 wedge group, n=18 control group) as per the statistical approach described in section 5.2.5, and thus the difference observed between the baseline and three months columns may not equate to those in the mean differences column. Variable KOOS (0-100) Pain Wedged Insole Mean (SD) Baseline (n=19) 51.5 (17.1) 3 Months (n=15) 56.6 (13.1) Baseline (n=19) 55.4 (13.5) Control Mean (SD) 3 Months (n=18) 55.6 (16.7) Mean (SD) Difference (3 months - Baseline) Wedged Insole Control (n=15) (n=18) 6.6 (17.7) 1.0 (12.8) Between Groups Mean Difference (95% CI) (Control Insole) -5.6 (-16.5 to 5.3) Symptoms 55.5 (16.7) 61.9 (13.5) 57.0 (14.5) 58.5 (16.0) 5.7 (17.3) 1.4 (15.1) -4.4 (-15.8 to 7.1) ADL 60.1 (16.2) 64.0 (15.2) 70.2 (14.5) 64.0 (17.8) 5.9 (14.9) -5.4 (16.8) (-22.6 to 0.1)* Sport/Rec 45.0 (27.5) 44.0 (18.7) 36.5 (22.9) 39.9 (23.7) -0.7 (28.2) 4.6 (23.3) 5.3 (-13.0 to 23.6) QOL 37.2 (18.1) 42.1 (14.3) 36.5 (15.2) 36.1 (17.1) 6.7 (14.5) 0.3 (17.2) -6.3 (-17.7 to 5.1) PASE, points (80.2) (77.9) (112.5) (93.9) 3.4 (70.4) (77.3) (-84.7 to 21.2) UCLA, 1-10 (1.4) (1.5) (2.1) (2.0) *p=0.053; this difference is likely due to the borderline difference at baseline. 0.3 (1.6) -0.7 (1.4) -1.0 (-2.1 to 0.1) 84

100 5.3.4 Co-interventions, Adherence & Adverse Events The two study groups were well matched in terms of co-intervention use at baseline (F(4,33)=0.90, p=0.476) and in terms of co-intervention use and adherence at follow-up (F(6,25)=1.07, p=0.408) (Table 5.4). However, the wedged insole group was found to experience significantly more new injuries than the control group (X 2 (1)=5.16, p=0.023) (Table 5.5). 85

101 Table 5.4. Co-intervention use and adherence characteristics are shown for both study groups. Variable Co-Intervention (days/week) Wedged Insole Baseline (n=19) 3 Months (n=15) Baseline (n=19) Control 3 Months (n=18) NSAID/Acetaminophen 3.2 (2.9) 3.2 (3.0) 3.1 (2.8) 2.9 (2.8) Physiotherapy/Targeted Exercise 1.9 (2.6) 1.5 (2.2) 0.9 (2.2) 0.6 (1.8) Compression/Tensor brace 0.8 (2.2) 1.3 (2.6) 0.3 (0.8) 0.2 (0.7) Narcotic medication 0.1 (0.3) 0.1 (0.5) 0 (0) 0.8 (2.0) Unloader brace 0 (0) 0 (0) 0 (0) 0 (0) Adherence Use of assigned footwear (days/week) (2.4) (3.3) Use of assigned footwear (hours/day) (4.0) (4.9) 86

102 Table 5.5. The new injuries reported throughout the 3 month study are shown as total number of participants experiencing at least one new injury. The specific injuries experienced are documented for each group; however, since some participants experienced more than one side effect, totals for the specific side effects do not equate to the total number of participants experiencing a side effect. Side Effects Wedged Insole (n=19) Control (n=19) Total, No. of patients 13 6 Side Effect Types Foot Tenderness/Pain 9 5 Fall on or Twist Knee 3 0 Plantar Pain 2 0 Leg Pain 2 0 Back Pain 2 0 Ankle Tenderness/Pain 1 1 Foot Blistering

103 5.4 Discussion The purpose of this study was to evaluate the effectiveness of KAM reduction on clinical outcomes in individuals with medial knee osteoarthritis. Based on existing literature, it was hypothesized that (1) pain reduction over three months would be associated with individuals experiencing reduced KAMs, and (2) pain reduction would have a dose-response relationship to KAMs, where larger KAM reductions would result in larger pain reductions. The results from the present study cannot support either of these hypotheses, as no differences were observed between the wedged insole and control groups in terms of clinical outcomes, and none of the knee loading variables were associated with change in KOOS pain score over three months. While the study recruited more individuals than originally planned for baseline testing, the exclusion rate based on biomechanical ineligibility was far greater than anticipated, resulting in sample sizes lower than originally planned based on sample size calculations. However, given the sample sizes for each group were near the originally desired sample size, and the statistical results of this study highlight the clear lack of significant differences between groups or even within groups, the data presented here would be expected to be similar even with larger sample sizes. This study is the first to report that KAM reduction does not appear to contribute to a greater therapeutic effect than a control group with short-term follow-up. Frontal-plane knee joint moments have been studied for decades in regards to their relation to knee OA development and progression, consistently showing that individuals with knee OA experience larger KAMs than healthy individuals, and that larger KAMs are associated with more severe disease. 6,9,30,106 However, studies attempting to reduce KAMs for individuals with knee OA have shown mixed clinical outcomes. Specific to footwear 88

104 orthotics, some studies noted a beneficial clinical effect of wedges, 15,69 while others found no beneficial effect. 13,21 A recent systematic review has suggested that part of this discrepancy may be due to the fact that some trials did not use a sham orthotic in their control group, which could result in the interpretation of placebo effects as actual insole effects in the experimental group. 14 Other authors have highlighted that another aspect contributing to the disparity in results across trials may be that many individuals receiving an orthotic experience an unintended increase in KAMs, potentially washing out actual effects of KAM reduction. 23 In the present trial, no clinical difference was observed between study groups at three months follow-up, and no relationship between KAM reduction and change in pain over three months, even with a control group who did not receive a sham orthotic. These results are in agreement with those of Jones et al., 68 who assessed the relationship between KAM reduction and pain at baseline; however, these authors had no follow-up intervention data. While regression analyses found no association between change in KAM and change in pain, the results from the current study do not necessarily discount the potential long-term importance of knee joint load reductions for individuals with knee OA. As mentioned previously, there is research to suggest that altered joint loading may contribute to osteoarthritis, both in human and animal models, 9,30,42,107 and thus load reduction may serve a role in disease prevention. In terms of joint pathology, Bennell et al., 37 have shown that the amount of medial tibial cartilage volume loss over one year for individuals with knee OA was positively associated with magnitude of the KAAI during walking. Additionally, it has been shown that KAM magnitude predicts radiographic disease progression over six years. 8 From a clinical perspective, Messier et al., 33 have shown that 89

105 individuals with knee OA who reduce their body mass by greater than 10% through diet and/or exercise experience greater reductions in knee joint compressive forces and greater reductions in pain over 18 months compared to those who lost less body mass. These studies highlight a potential beneficial effect of load reduction in the long term ( 1 year follow-up) for OA management. Therefore, it may be possible that beneficial effects of load reduction are not apparent in short-term follow-ups, such as the three month followup described in this study, or in other studies with exclusively baseline assessments. 68 This notion may explain why no association was found between load reduction and change in pain in the current study. Consequently, further studies assessing the effects of confirmed load reduction on long term outcomes, whether it be clinical improvement, or structural maintenance of the joint should be considered. Another important finding in the current study was that reductions in KOOS pain over three months for the wedged insole group were associated with a more severe baseline pain (lower KOOS), and a reduced physical activity level (reduced PASE). It should be highlighted that only 5/19 individuals randomized to the wedged insole group experienced a clinically relevant improvement in KOOS score; however, the identified relationship between change in pain and baseline pain is still important to consider. Specifically, in a meta-analysis, it was shown that placebo effects in OA trials are often largest for individuals whose baseline pain is more severe 108 in essence, these patients have more to gain. Thus, the finding that changes in pain (which were mostly non-clinically meaningful) were related to baseline pain score is suggestive of a placebo effect of the insoles in the short term. Since the control group did not receive a sham orthotic, it would therefore make sense that a similar effect was not seen in the control group. In regards to the association 90

106 with change in PASE score, it has been highlighted recently that physical activity must always be accounted for in OA trials. 109 One reason for this is that physical activity may induce positive psychological and/or physiological changes that ultimately alter the processing of pain. 110 Another possibility is that with altered physical activity levels, total cumulative joint loading may be altered, which could contribute towards a lack of association between KAM reduction and pain change. 48 Nonetheless, it is believed that placebo effects likely account for the majority of the variance in change in KOOS pain score in the current study since baseline pain was the larger contributor to the regression model. In the present study, co-intervention use was unchanged over the duration of the trial for either intervention group. Previous trials have found that in some cases, pain may not change, but a reduction in co-intervention use may occur, which ultimately may indicate a therapeutic effect. 21 Contrary to this, the present trial found that wedged insoles were significantly associated with more adverse events compared to the control group. It should be noted that these findings may be associated with the fact that only the wedged insole group actually received a new intervention. While these adverse events were generally no more than foot or leg discomfort or cramping, three individuals did drop-out from the trial due to foot discomfort. There are a number of limitations to this study. Primarily, the study included a relatively short follow-up, and thus the long-term effects of load reduction using wedged insoles remain unknown. While a relatively low sample size was utilized, the statistical results from the data strongly suggest that no clinically meaningful association between load reduction and change in pain exists in the short-term, and these data would not likely 91

107 be altered by larger sample sizes. Supporting this point, the results presented here are in agreement with a larger study assessing baseline changes in pain from reduced KAMs. 68 Additionally, since biomechanical variables were not assessed at follow-up, it was assumed that the biomechanical changes induced in the laboratory at baseline would be maintained throughout the trial duration and during prolonged activity, as has been shown to be the case in other studies. 95,111 It remains a possibility that the effects of the wedged insoles changed over time, or that the individual s own shoes had different effects over time with shoe wear. Lastly, while medial knee OA was the predominant compartment of disease in the study population, some participants also had OA in other compartments, which could potentially affect clinical responses to altered knee mechanics and wedged insoles. 5.5 Conclusion This was the first study to demonstrate the effects of reduced KAMs via wedged insoles on clinical outcomes over time. After 3-months follow-up, no differences were observed between the wedged insole group and control group in terms of pain, physical activity, or co-intervention use. Additionally, reduced KAMs were not found to be associated with reduced pain over the duration of the study. Instead, baseline pain and change in physical activity over three months were associated with change in pain. Taken together, these results suggest a mild, non-clinically meaningful placebo effect of wedged insoles, and no loading-related clinical changes induced by the insoles over a 3-month period. 92

108 CHAPTER SIX: SUMMARY 6.1 Overview of Rationale Knee osteoarthritis is a highly prevalent disease in Canada, 1 contributing to significant disability and health care-related costs. 2 Abnormally high frontal-plane knee joint loading, as measured by the knee adduction moment (KAM), has long been viewed as a key contributor to disease development and progression, 6,30 and consequently reductions to these loads have been a priority of conservative management. Wedged footwear insoles are commonly applied for this purpose; however, clinical studies evaluating the effectiveness of wedged insoles have shown mixed results in terms of clinical efficacy. 13,15,21 While the reasons for these mixed results may be multifactorial, two major issues are well established: (1) it is imperative that a proper control condition is utilized, and (2) effort must be made to ensure that all participants receiving a wedged insole actually experience reduced KAMs. The rationale behind the first point is based on the fact that differing control conditions seem to yield differing clinical conclusions, 14 and certain sham orthoses can inadvertently affect foot pressure beneath the foot, which may affect KAMs. 23,70 Regarding the second point, it has been found that many patients receiving a wedged insole actually experience an unintended increase in KAMs. 16 Unfortunately, it was not known how to predict which patients may experience KAM reductions, and which ones will not, resulting in patients all being assigned the same wedge type in clinical studies. For patients experiencing increased KAMs from the insoles, this 93

109 may have contributed to disease and symptom worsening, which could bias study outcomes. Consequently, the true effect of reduced KAMs has never been evaluated. There were three primary research objectives that this thesis sought to investigate. These objectives were as follows: (4) Identify the footwear/insole/orthotic condition that should be utilized as a control condition in randomized controlled trials aiming to assess the effects of reduced KAMs on clinical outcomes. (5) Identify a method that can be used to predict the expected change in KAM magnitude from a wedged insole intervention. (6) Identify the influence of reduced KAMs on clinical outcomes for individuals with medial knee OA. In the following section, a summary of results obtained from each of three studies (Chapters 3-5) that address these objectives will be provided, along with an overview of the result interpretations and limitations. 6.2 Results, Interpretations and Limitations Chapter 3 Control Conditions In this study, four different commonly utilized control conditions were evaluated in terms of their unintended biomechanical effects. The four conditions were as follows: (1) participant s own shoe, (2) participant s own shoe with a flat 3mm insole applied bilaterally, (3) a standardized running shoe, and (4) a standardized running shoe with a flat 94

110 3mm insole applied bilaterally. Participants were asked to walk in a biomechanics laboratory with each condition, and a variety of lower extremity biomechanical variables were evaluated, including the knee adduction moment. When compared to the participants own footwear, all other conditions resulted in a significant proportion of participants experiencing KAM changes greater than ±10% of the value measured while wearing their own footwear. Given the objective of a control condition is to ensure participants allocated to this condition do not experience a change to the primary variable under study, 23 it was concluded that the most suitable control condition for studies evaluating the clinical effects of reduced KAMs should utilize the participant s own footwear as the control condition and not utilize a sham orthotic, or standardized footwear. One limitation of utilizing the participant s own footwear as a control group is that intervention placebo effects will not be accounted for in study analysis, given that control participants know they have not received an intervention. 92 However, this method does ensure the participant s KAM remains as close as possible to what the individual normally experiences, which is necessary if changes to KAMs is the primary variable under study. A general limitation of this specific study is that since only healthy individuals were utilized, it was assumed that similar results would be found for knee osteoarthritis individuals. Since healthy individuals and knee osteoarthritis individuals show similar responses to wedged insoles, 10,17,64,66 this assumption seemed reasonable. Additionally, the study selected a somewhat arbitrary cut-off value for biomechanical importance of ±10% of the magnitude observed with the participants own footwear. This value was chosen since 95

111 it is approximately the magnitude of change induced by actual wedged insoles; 16,19 however, it is currently not known if a biomechanical change of ±10% is clinically relevant Chapter 4 Wedge Prediction In this study, attempt was made to identify a possible method for predicting an individual s KAM response to lateral and medial wedge insoles. Both healthy and knee osteoarthritic individuals participated in the study, completing trials of walking, and of single-step movements in a gait laboratory. It was identified that the two-dimensional mediolateral position changes (as wedge neutral) to the knee joint center, leg center of mass, ankle joint center and foot center of mass could be used to predict the KAM change that a wedged insole would induce. Using this procedure for both lateral and medial wedges, between 80-89% of participants could be correctly classified into either an insole recommendation to reduce KAMs, or a no insole recommendation to prevent KAM increase. While this method is still in its early stages of development, this proof-of-concept study provides the framework for developing an automated, low-cost system for predicting the effects of wedged insoles in non-research lab settings, and thus has the potential to be a highly utilized technology. The largest source of error in this study was that marker positions at the knee and ankle were reapplied with each new footwear condition rather than remaining fixed to the participant throughout all conditions and trials. Marker reapplication was always done by the same researcher, which helped to mitigate any placement error, but slight changes in marker positions are still possible, which could introduce error to the joint center 96

112 calculations. This may have weakened the strength of the relationship observed between the predictor variables and the change in KAM during walking Chapter 5 Three-Month Randomized Trial of Reduced KAMs In this study, individuals with knee osteoarthritis were randomly allocated either to a control group, where no footwear intervention was given, or an insole group, where a wedged insole was given that was verified to reduce KAMs for each participant. All participants were evaluated at baseline in terms of their KAMs, symptoms using the Knee Injury and Osteoarthritis Outcome Score (KOOS) and physical activity using the Physical Activity Scale for the Elderly (PASE). Participants were monitored for three months, and repeat KOOS and PASE surveys were administered, as well as a co-intervention survey, and new injury survey. It was found that there was no difference in terms of KOOS pain change over three months between study groups, and it was also found that there was no association between change in KAM magnitude and change in KOOS pain over three months within the wedged insole group. Interestingly, it was found that improvements in KOOS pain for the wedged insole group were associated with worse baseline pain, and a change in physical activity levels over three months. These results highlight that reduced KAMs do not appear to provide any clinical benefit over a short follow-up period of three months, and any changes observed within the wedged insole group were likely due to placebo effects. 108 This study was limited by a low sample size compared to other, larger randomized trials for knee osteoarthritis; 13,33 however, given the very clear lack of relationship between knee joint loading and change in pain over three months, and lack of differences between 97

113 study groups, our results would be unlikely to change even with larger sample sizes. More importantly, the study would have benefitted from a longer follow-up duration to assess whether clinically meaningful effects of reduced KAMs only present at longer follow-ups. Additionally, while medial knee OA was the predominant compartment of disease in the study population, some participants also had disease in other compartments, which could affect clinical responses to altered knee mechanics and wedged insoles. 6.3 Future Directions This thesis has highlighted the effects of different control conditions, and how this might be important in designing clinical studies for individuals with knee osteoarthritis, it has identified a potential method that may be broadly implemented to predict the expected KAM response induced by wedged insoles, and has provided evidence that KAM reduction does not appear to offer a clinical benefit in the short-term. While these data contribute to our general knowledge of the mechanics of wedged insoles, gait, and knee osteoarthritis, there is still much work to be done on these topics. Firstly, it would be advantageous to repeat the control conditions study in a knee osteoarthritis study population. As mentioned previously, it is expected that similar results would be found as the present study; however, this assumption could have very important consequences if incorrect. In future randomized trials, it may be beneficial to include two control groups: one group who receive no intervention, and a second who receive a flat sham orthotic. While this would require a much larger sample size, it could be of benefit in that it would allow for assessment of both biomechanical and placebo effects of insoles. 98

114 In regards to prediction of KAM response with wedged insoles, the critical next step will be to implement this approach using less sophisticated and expensive equipment and determine if similar predictive capabilities can be achieved. Specifically, predictor variables used were two-dimensional, yet these data were collected in three-dimensions using highly sophisticated equipment, and so it remains unknown as to whether similar predictive ability can be achieved under a more simple 2D data collection. Given that taking a single step is a relatively low speed movement, the likelihood of finding success when adapting the method to simpler data collection systems seems promising. Next, effort must be made to automate the system under a user-friendly interface such that it can be highly accessible to both clinicians and the general public. The method would also benefit from including data of more participants to further refine and improve the regression equations, especially in the case of medial wedge insoles where predictive capacity was lower than laterally wedged insoles in the method s current form. Finally, it is apparent that further research is needed to fully understand the influence of reduced KAMs on clinical outcomes, as the present thesis only scratches the surface of this area by providing a short-term follow-up, and only assessing pain. In the future, it would be important to repeat the study with longer trial durations, and also including structural measures of knee joint capacity such as radiographic disease progression, 8 or tibial cartilage volume by magnetic resonance imaging. 13 The reason for this is that while reduced KAMs may or may not have an effect on pain, it is possible that reduced KAMs could be important for joint preservation, which would only be captured experimentally with long trial durations. In addition, the time course of knee osteoarthritis must be further described from a mechanical standpoint current evidence shows that 99

115 individuals with knee OA experience large KAMs; 30 however, it remains unclear if early intervention with reduced KAMs has the potential to prevent OA. Certainly, this would necessitate a very large sample size and long follow-up duration. In the interim, while this study cannot offer support in favor of using wedged insoles for conservative management of knee osteoarthritis, it also cannot provide evidence of clear contraindication to wedged insole use, aside from an increased risk of side effects such as foot tenderness and discomfort, among other relatively low-risk side effects. Clinically, when considering alternative high-risk or invasive therapies such as joint replacement, unloader braces, or viscosupplementation, intervention with wedged insoles seems like a reasonable option for patients to consider, but clinicians should warn their patients that there is a risk of increased discomfort, as well as a high probability that no clinical benefit will be noted, at least in the short-term. 100

116 REFERENCES 1. Public Health Agency of Canada. Arthritis in Canada: an ongoing challenge. Health Canada, Ottawa; The Arthritis Society of Canada. Arthritis Facts & Figures p The Centers for Disease Control and Prevention. Osteoarthritis. May 16, Felson DT. Clinical practice. Osteoarthritis of the knee. N Engl J Med 2006; 354(8): Bennell KL, Hunter DJ, Hinman RS. Management of osteoarthritis of the knee. BMJ 2012; 345: e Reeves ND, Bowling FL. Conservative biomechanical strategies for knee osteoarthritis. Nat Rev Rheumatol 2011; 7(2): Kutzner I, Trepczynski A, Heller MO, Bergmann G. Knee adduction moment and medial contact force--facts about their correlation during gait. PLoS One 2013; 8(12): e Miyazaki T. Dynamic load at baseline can predict radiographic disease progression in medial compartment knee osteoarthritis. Ann Rheum Dis 2002; 61(7): Sharma L, Hurwitz DE, Thonar EJMA, et al. Knee adduction moment, serum hyaluronan level, and disease severity in medial tibiofemoral osteoarthritis. Arthritis Rheum 1998; 41(7):

117 10. Lewinson RT, Fukuchi CA, Worobets JT, Stefanyshyn DJ. The effects of wedged footwear on lower limb frontal plane biomechanics during running. Clin J Sport Med 2013; 23(3): Radzimski AO, Mundermann A, Sole G. Effect of footwear on the external knee adduction moment - A systematic review. Knee 2012; 19(3): Hinman RS, Bowles KA, Metcalf BB, Wrigley TV, Bennell KL. Lateral wedge insoles for medial knee osteoarthritis: effects on lower limb frontal plane biomechanics. Clin Biomech 2012; 27(1): Bennell KL, Bowles KA, Payne C, et al. Lateral wedge insoles for medial knee osteoarthritis: 12 month randomised controlled trial. BMJ 2011; 342: d Parkes MJ, Maricar N, Lunt M, et al. Lateral wedge insoles as a conservative treatment for pain in patients with medial knee osteoarthritis: a meta-analysis. JAMA 2013; 310(7): Erhart JC, Mundermann A, Elspas B, Giori NJ, Andriacchi TP. Changes in knee adduction moment, pain, and functionality with a variable-stiffness walking shoe after 6 months. J Orthop Res 2010; 28(7): Chapman GJ, Parkes MJ, Forsythe L, Felson DT, Jones RK. Ankle motion influences the external knee adduction moment and may predict who will respond to lateral wedge insoles?: an ancillary analysis from the SILK trial. Osteoarthritis Cartilage 2015; 23(8): Hinman RS, Payne C, Metcalf BR, Wrigley TV, Bennell KL. Lateral wedges in knee osteoarthritis: what are their immediate clinical and biomechanical effects and can these predict a three-month clinical outcome? Arthritis Rheum 2008; 59(3):

118 18. Hinman RS, Bowles KA, Payne C, Bennell KL. Effect of length on laterally-wedged insoles in knee osteoarthritis. Arthritis Rheum 2008; 59(1): Butler RJ, Marchesi S, Royer T, Davis IS. The effect of a subject-specific amount of lateral wedge on knee mechanics in patients with medial knee osteoarthritis. J Orthop Res 2007; 25(9): Kakihana W, Akai M, Nakazawa K, Naito K, Torii S. Inconsistent knee varus moment reduction caused by a lateral wedge in knee osteoarthritis. Am J Phys Med Rehabil 2007; 86(6): Pham T, Maillefert JF, Hudry C, et al. Laterally elevated wedged insoles in the treatment of medial knee osteoarthritis. A two-year prospective randomized controlled study. Osteoarthritis Cartilage 2004; 12(1): Toda Y, Tsukimura N. A 2-year follow-up of a study to compare the efficacy of lateral wedged insoles with subtalar strapping and in-shoe lateral wedged insoles in patients with varus deformity osteoarthritis of the knee. Osteoarthritis Cartilage 2006; 14(3): Lewinson RT, Stefanyshyn DJ. Losing control over control conditions in knee osteoarthritis orthotic research. Contemp Clin Trials 2015; 42: Duncan RC, Hay EM, Saklatvala J, Croft PR. Prevalence of radiographic osteoarthritis--it all depends on your point of view. Rheumatology 2006; 45(6): Wise BL, Niu J, Yang M, et al. Patterns of compartment involvement in tibiofemoral osteoarthritis in men and women and in whites and African Americans. Arthritis Care Res 2012; 64(6):

119 26. Friel NA, Chu CR. The role of ACL injury in the development of posttraumatic knee osteoarthritis. Clin Sports Med 2013; 32(1): Zhuo Q, Yang W, Chen J, Wang Y. Metabolic syndrome meets osteoarthritis. Nat Rev Rheumatol 2012; 8(12): Loughlin J. Genetic contribution to osteoarthritis development: current state of evidence. Curr Opin Rheumatol 2015; 27(3): Altman R, Alarcón G, Appelrouth D, et al. The American College of Rheumatology criteria for the classification and reporting of osteoarthritis of the hip. Arthritis Rheum 1991; 34(5): Andriacchi TP, Mündermann A, Smith RL, Alexander EJ, Dyrby CO, Koo S. A Framework for the in Vivo Pathomechanics of Osteoarthritis at the Knee. Ann Biomed Eng 2004; 32(3): Waller KA, Zhang LX, Elsaid KA, Fleming BC, Warman ML, Jay GD. Role of lubricin and boundary lubrication in the prevention of chondrocyte apoptosis. Proc Natl Acad Sci U S A 2013; 110(15): Wong BL, Kim SH, Antonacci JM, McIlwraith CW, Sah RL. Cartilage shear dynamics during tibio-femoral articulation: effect of acute joint injury and tribosupplementation on synovial fluid lubrication. Osteoarthritis Cartilage 2010; 18(3): Messier SP, Mihalko SL, Legault C, et al. Effects of intensive diet and exercise on knee joint loads, inflammation, and clinical outcomes among overweight and obese adults with knee osteoarthritis: the IDEA randomized clinical trial. JAMA 2013; 310(12):

120 34. Liu X, Zhang M. Redistribution of knee stress using laterally wedged insole intervention: Finite element analysis of knee-ankle-foot complex. Clin Biomech 2013; 28(1): Erhart JC, Dyrby CO, D'Lima DD, Colwell CW, Andriacchi TP. Changes in in vivo knee loading with a variable-stiffness intervention shoe correlate with changes in the knee adduction moment. J Orthop Res 2010; 28(12): Walter JP, D'Lima DD, Colwell CW, Jr., Fregly BJ. Decreased knee adduction moment does not guarantee decreased medial contact force during gait. J Orthop Res 2010; 28(10): Bennell KL, Bowles KA, Wang Y, Cicuttini F, Davies-Tuck M, Hinman RS. Higher dynamic medial knee load predicts greater cartilage loss over 12 months in medial knee osteoarthritis. Ann Rheum Dis 2011; 70(10): Park SK, Stefanyshyn DJ. Greater Q angle may not be a risk factor of patellofemoral pain syndrome. Clin Biomech 2011; 26(4): Sharma L, Song J, Felson DT, Cahue S, Shamiyeh E, Dunop DD. The Role of Knee Alignment in Disease Progression and Functional Decline in Knee Osteoarthritis. JAMA 2001; 286(2): Foroughi N, Smith RM, Lange AK, Baker MK, Fiatarone Singh MA, Vanwanseele B. Dynamic alignment and its association with knee adduction moment in medial knee osteoarthritis. Knee 2010; 17(3): Fukutani N, Iijima H, Fukumoto T, et al. Association Between Varus Thrust and "Pain and Stiffness" and "Activities of Daily Living" in Patients With Medial Knee Osteoarthritis. Phys Ther

121 42. Rehan Youssef A, Longino D, Seerattan R, Leonard T, Herzog W. Muscle weakness causes joint degeneration in rabbits. Osteoarthritis Cartilage 2009; 17(9): Ruhdorfer AS, Dannhauer T, Wirth W, et al. Thigh muscle cross-sectional areas and strength in knees with early vs knees without radiographic knee osteoarthritis: a between-knee, within-person comparison. Osteoarthritis Cartilage 2014; 22(10): Foroughi N, Smith RM, Lange AK, Baker MK, Fiatarone Singh MA, Vanwanseele B. Lower limb muscle strengthening does not change frontal plane moments in women with knee osteoarthritis: A randomized controlled trial. Clin Biomech 2011; 26(2): Hinman RS, Hunt MA, Creaby MW, Wrigley TV, McManus FJ, Bennell KL. Hip muscle weakness in individuals with medial knee osteoarthritis. Arthritis Care Res 2010; 62(8): Lewinson RT, Worobets JT, Stefanyshyn DJ. The relationship between maximal hip abductor strength and resultant loading at the knee during walking. Proc Inst Mech Eng H 2014; 228(12): Kean CO, Bennell KL, Wrigley TV, Hinman RS. Relationship between hip abductor strength and external hip and knee adduction moments in medial knee osteoarthritis. Clin Biomech 2015; 30(3): Maly MR. Abnormal and cumulative loading in knee osteoarthritis. Curr Opin Rheumatol 2008; 20(5):

122 49. Kito N, Shinkoda K, Yamasaki T, et al. Contribution of knee adduction moment impulse to pain and disability in Japanese women with medial knee osteoarthritis. Clin Biomech 2010; 25(9): Winter DA. Biomechanics and motor control of human movement. 4th ed. Hoboken, NJ: John WIley & Sons; Robertson DGE, Caldwell GE, Hamill J, Kamen G, Whittlesey SN. Research methods in biomechanics. 1st ed. Champaign, IL: Human Kinetics; Nigg BM, Herzog W. Biomechanics of the musculo-skeletal system. 3rd ed. Hoboken, NJ: John Wiley & Sons; Dempster WD. Space requirements of the seated operator: geometrical, kinematic and mechanical aspects of the body with special reference to the limbs. Wright-Patterson Airforce Base, Ohio: Aero Medical Laboratory; Clauser CE, McConville JT, Young JW. Weight, volume, and center of mass of segments of the human body. Wright-Patterson Airforce Base, Ohio: Aerospace Medical Research Laboratory; Cole GK, Nigg BM, Ronsky JL, Yeadon MR. Application of the Joint Coordinate System to Three-Dimensional Joint Attitude and Movement Representation: A Standardization Proposal. J Biomech Eng 1993; 115(4A): Grood ES, Suntay WJ. A Joint Coordinate System for the Clinical Description of Three-Dimensional Motions: Application to the Knee. J Biomech Eng 1983; 105(2): Crowninshield RD, Brand RA. A physiologically based criterion of muscle force prediction in locomotion. J Biomech 1981; 14(11):

123 58. Adouni M, Shirazi-Adl A. Partitioning of knee joint internal forces in gait is dictated by the knee adduction angle and not by the knee adduction moment. J Biomech 2014; 47(7): Stefanyshyn DJ, Stergiou P, Lun VM, Meeuwisse WH, Worobets JT. Knee angular impulse as a predictor of patellofemoral pain in runners. Am J Sports Med 2006; 34(11): Kadaba MP, Ramakrishnan HK, Wootten ME. Measurement of lower extremity kinematics during level walking. J Orthop Res 1990; 8(3): Teoh JC, Low JH, Lim YB, et al. Investigation of the biomechanical effect of variable stiffness shoe on external knee adduction moment in various dynamic exercises. J Foot Ankle Res 2013; 6(1): Lewinson RT, Worobets JT, Stefanyshyn DJ. Calculation of external knee adduction moments: A comparison of an inverse dynamics approach and a simplified lever-arm approach. Knee 2015; 22(4): Erhart JC, Mundermann A, Mundermann L, Andriacchi TP. Predicting changes in knee adduction moment due to load-altering interventions from pressure distribution at the foot in healthy subjects. J Biomech 2008; 41(14): Schmalz T, Blumentritt S, Drewitz H, Freslier M. The influence of sole wedges on frontal plane knee kinetics, in isolation and in combination with representative rigid and semi-rigid ankle-foot-orthoses. Clin Biomech 2006; 21(6): Kakihana W, Akai M, Nakazawa K, Takashima T, Naito K, Torii S. Effects of Laterally Wedged Insoles on Knee and Subtalar Joint Moments. Arch Phys Med Rehabil 2005; 86(7):

124 66. Nigg BM, Stergiou P, Cole G, Stefanyshyn D, Mundermann A, Humble N. Effect of shoe inserts on kinematics, center of pressure, and leg joint moments during running. Med Sci Sports Exerc 2003; 35(2): Zhang W, Moskowitz RW, Nuki G, et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part II: OARSI evidence-based, expert consensus guidelines. Osteoarthritis Cartilage 2008; 16(2): Jones RK, Chapman GJ, Forsythe L, Parkes MJ, Felson DT. The relationship between reductions in knee loading and immediate pain response whilst wearing lateral wedged insoles in knee osteoarthritis. J Orthop Res 2014; 32(9): Toda Y, Tsukimura N, Segal N. An optimal duration of daily wear for an insole with subtalar strapping in patients with varus deformity osteoarthritis of the knee. Osteoarthritis Cartilage 2005; 13(4): McCormick CJ, Bonanno DR, Landorf KB. The effect of customised and sham foot orthoses on plantar pressures. J Foot Ankle Res 2013; 6: Fantini Pagani CH, Hinrichs M, Bruggemann GP. Kinetic and kinematic changes with the use of valgus knee brace and lateral wedge insoles in patients with medial knee osteoarthritis. J Orthop Res 2012; 30(7): Jones RK, Nester CJ, Richards JD, et al. A comparison of the biomechanical effects of valgus knee braces and lateral wedged insoles in patients with knee osteoarthritis. Gait Posture 2013; 37(3): Moyer RF, Birmingham TB, Bryant DM, Giffin JR, Marriott KA, Leitch KM. Biomechanical effects of valgus knee bracing: a systematic review and meta-analysis. Osteoarthritis Cartilage 2015; 23(2):

125 74. Arroll B, Goodyear-Smith F. Corticosteroid injections for osteoarthritis of the knee: meta-analysis. BMJ 2004; 328(7444): Bellamy N, Campbell J, Robinson V, Gee T, Bourne R, Wells G. Intraarticular corticosteroid for treatment of osteoarthritis of the knee. Cochrane Database Syst Rev 2006; (2): CD Hunter DJ. Viscosupplementation for osteoarthritis of the knee. N Engl J Med 2015; 372(11): Rutjes AW, Juni P, da Costa BR, Trelle S, Nuesch E, Reichenbach S. Viscosupplementation for osteoarthritis of the knee: a systematic review and metaanalysis. Ann Intern Med 2012; 157(3): Briem K, Axe MJ, Snyder-Mackler L. Medial knee joint loading increases in those who respond to hyaluronan injection for medial knee osteoarthritis. J Orthop Res 2009; 27(11): Wang SY, Olson-Kellogg B, Shamliyan TA, Choi JY, Ramakrishnan R, Kane RL. Physical therapy interventions for knee pain secondary to osteoarthritis: a systematic review. Ann Intern Med 2012; 157(9): Bennell KL, Buchbinder R, Hinman RS. Physical therapies in the management of osteoarthritis: current state of the evidence. Curr Opin Rheumatol 2015; 27(3): Doherty C, Bleakley C, Hertel J, Caulfield B, Ryan J, Delahunt E. Lower extremity function during gait in participants with first time acute lateral ankle sprain compared to controls. J Electromyogr Kinesiol 2015; 25(1):

126 82. Zadpoor AA, Nikooyan AA. The relationship between lower-extremity stress fractures and the ground reaction force: a systematic review. Clin Biomech 2011; 26(1): Mündermann A, Nigg BM, Neil Humble R, Stefanyshyn DJ. Foot orthotics affect lower extremity kinematics and kinetics during running. Clin Biomech 2003; 18(3): Barton CJ, Munteanu SE, Menz HB, Crossley KM. The efficacy of foot orthoses in the treatment of individuals with patellofemoral pain syndrome: a systematic review. Sports Med 2010; 40(5): Ferber R, McClay Davis I, Williams DS, Laughton C. A comparison of within- and between-day reliability of discrete 3D lower extremity variables in runners. J Orthop Res 2002; 20(6): Asay JL, Boyer KA, Andriacchi TP. Repeatability of gait analysis for measuring knee osteoarthritis pain in patients with severe chronic pain. J Orthop Res 2013; 31(7): Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, Cochran GV. Repeatability of kinematic, kinetic, and electromyographic data in normal adult gait. J Orthop Res 1989; 7(6): Kerrigan DC, Lelas JL, Goggins J, Merriman GJ, Kaplan RJ, Felson DT. Effectiveness of a lateral-wedge insole on knee varus torque in patients with knee osteoarthritis. Arch Phys Med Rehabil 2002; 83(7):

127 89. Franz JR, Dicharry J, Riley PO, Jackson K, Wilder RP, Kerrigan DC. The influence of arch supports on knee torques relevant to knee osteoarthritis. Med Sci Sports Exerc 2008; 40(5): Miller RH, Hamill J. Computer simulation of the effects of shoe cushioning on internal and external loading during running impacts. Comput Methods Biomech Biomed Eng 2009; 12(4): Kakihana W, Akai M, Yamasaki N, Takashima T, Nakazawa K. Changes of Joint Moments in the Gait of Normal Subjects Wearing Laterally Wedged Insoles. Am J Phys Med Rehabil 2004; 83(4): Bonanno DR, Landorf KB, Murley GS, Menz HB. Selecting control interventions for use in orthotic trials: The methodological benefits of sham orthoses. Contemp Clin Trials 2015; 42: Lewinson RT, Collins KH, Vallerand IA, et al. Reduced knee joint loading with lateral and medial wedge insoles for management of knee osteoarthritis: a protocol for a randomized controlled trial. BMC Musculoskelet Disord 2014; 15: Altman R, Asch E, Bloch D, et al. Development of criteria for the classification and reporting of osteoarthritis: Classification of osteoarthritis of the knee. Arthritis Rheum 1986; 29(8): Lewinson RT, Worobets JT, Stefanyshyn DJ. Knee abduction angular impulses during prolonged running with wedged insoles. Proc Inst Mech Eng H 2013; 227(7): Allen DM. The Relationship Between Variable Selection and Data Agumentation and a Method for Prediction. Technometrics 1974; 16(1):

128 97. Metcalfe AJ, Stewart C, Postans N, Dodds AL, Holt CA, Roberts AP. The effect of osteoarthritis of the knee on the biomechanics of other joints in the lower limbs. Bone Joint J 2013; 95-B(3): Kellgren JH, Lawrence JS. Radiological Assessment of Osteo-Arthrosis. Ann Rheum Dis 1957; 16(4): Roos EM, Lohmander LS. The Knee injury and Osteoarthritis Outcome Score (KOOS): from joint injury to osteoarthritis. Health Qual Life Outcomes 2003; 1: Washburn RA, Smith KW, Jette AM, Janney CA. The physical activity scale for the elderly (PASE): Development and evaluation. J Clin Epidemiol 1993; 46(2): Zahiri CA, Schmalzried TP, Szuszczewicz ES, Amstutz HC. Assessing activity in joint replacement patients. J Arthroplasty 1998; 13(8): Collins NJ, Misra D, Felson DT, Crossley KM, Roos EM. Measures of knee function: International Knee Documentation Committee (IKDC) Subjective Knee Evaluation Form, Knee Injury and Osteoarthritis Outcome Score (KOOS), Knee Injury and Osteoarthritis Outcome Score Physical Function Short Form (KOOS-PS), Knee Outcome Survey Activities of Daily Living Scale (KOS-ADL), Lysholm Knee Scoring Scale, Oxford Knee Score (OKS), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), Activity Rating Scale (ARS), and Tegner Activity Score (TAS). Arthritis Care Res 2011; 63 Suppl 11: S Henriksen M, Klokker L, Graven-Nielsen T, et al. Association of exercise therapy and reduction of pain sensitivity in patients with knee osteoarthritis: a randomized controlled trial. Arthritis Care Res 2014; 66(12):

129 104. Armijo-Olivo S, Warren S, Magee D. Intention to treat analysis, compliance, dropouts and how to deal with missing data in clinical research: a review. Phys Ther Rev 2009; 14(1): Vickers AJ, Altman DG. Statistics Notes: Analysing controlled trials with baseline and follow up measurements. BMJ 2001; 323(7321): Butler RJ, Barrios JA, Royer T, Davis IS. Frontal-plane gait mechanics in people with medial knee osteoarthritis are different from those in people with lateral knee osteoarthritis. Phys Ther 2011; 91(8): Roemhildt ML, Coughlin KM, Peura GD, et al. Effects of increased chronic loading on articular cartilage material properties in the lapine tibio-femoral joint. J Biomech 2010; 43(12): Zhang W, Robertson J, Jones AC, Dieppe PA, Doherty M. The placebo effect and its determinants in osteoarthritis: meta-analysis of randomised controlled trials. Ann Rheum Dis 2008; 67(12): Lo GH, McAlindon TE, Hawker GA, et al. Symptom Assessment in Knee Osteoarthritis Needs to Account for Physical Activity Level. Arthritis Rheumatol 2015; 67(11): Scheef L, Jankowski J, Daamen M, et al. An fmri study on the acute effects of exercise on pain processing in trained athletes. Pain 2012; 153(8): Hinman RS, Bowles KA, Bennell KL. Laterally wedged insoles in knee osteoarthritis: do biomechanical effects decline after one month of wear? BMC Musculoskelet Disord 2009; 10:

130 APPENDIX A: BASELINE DATA COLLECTION FORM 115

131 116

132 117

133 APPENDIX B: CHAPTER FIVE RCT PROTOCOL 118

134 119

135 120

136 121

137 122

138 123

139 124

140 125