Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Agreement and validity of observational risk screening guidelines in evaluating ACL injury risk factors Ekegren, Christina Louise 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2008_spring_ekegren_christina_louise.pdf [ 29.17MB ]
JSON: 24-1.0066595.json
JSON-LD: 24-1.0066595-ld.json
RDF/XML (Pretty): 24-1.0066595-rdf.xml
RDF/JSON: 24-1.0066595-rdf.json
Turtle: 24-1.0066595-turtle.txt
N-Triples: 24-1.0066595-rdf-ntriples.txt
Original Record: 24-1.0066595-source.json
Full Text

Full Text

AGREEMENT AND VALIDITY OF OBSERVATIONAL RISK SCREENING GUIDELINES IN EVALUATING ACL INJURY RISK FACTORS by CHRISTINA LOUISE EKEGREN B.Physiotherapy (Hons), The University of Melbourne, 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Rehabilitation Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA December 2007 © Christina Louise Ekegren, 2007  ABSTRACT Study Design: Methodological study. Objectives: To examine the agreement and validity of using observational risk screening guidelines to evaluate ACL injury risk factors. Background: Post-pubescent females have an increased risk of anterior cruciate ligament (ACL) injury compared with their male counterparts partly due to their high-risk landing and cutting strategies. There are currently no scientifically-tested methods to screen for these high risk strategies in the clinic or on the field. Methods and Measures: Three physiotherapists used observational risk screening guidelines to rate the neuromuscular characteristics of 40 adolescent female soccer players. Drop jumps were rated as high risk or low risk based on the degree of knee abduction. Side hops and side cuts were rated on the degree of lower limb 'reaching'. Ratings were evaluated for intrarater and interrater agreement using kappa coefficients. 3D motion analysis was used as a gold standard for determining the validity of ratings. Results: Acceptable intrarater and interrater agreement (k ^0.61) were attained for the drop jump and the side hop, with kappa coefficients ranging from 0.64 to 0.94. Acceptable sensitivity ( ^0.80) was attained for the side hop and the side cut, with values ranging from 0.88 to 1.00. Acceptable specificity (^0.50) was attained for the drop jump, with values ranging from 0.64 to 0.72. Conclusion: Observational risk screening is a practical and cost-effective method of screening for ACL injury risk. Based on levels of agreement and sensitivity, the side hop appears to be a suitable screening task. Agreement was acceptable for the drop jump but its validity needs further investigation. Key Words: female, knee, movement analysis, neuromuscular, soccer 11  ^  TABLE OF CONTENTS  ABSTRACT ^ LIST OF TABLES ^  vii  LIST OF FIGURES ^  viii  LIST OF ABBREVIATIONS ^  ix  GLOSSARY ^ ACKNOWLEDGMENTS ^  xiii  1 INTRODUCTION ^ 1.1^ACL INJURIES - THE SCOPE OF THE PROBLEM ^  1 1  1.1.1^Non Contact ACL Injuries ^  2  1.1.2^ACL Injuries and the Female Athlete ^  2  1.2 ACL ANATOMY ^  3  1.3 ACL INJURY MECHANISMS ^  4  1.4 EFFECT OF THESE MECHANISMS ON THE ACL ^  4  1.4.1^Dynamic Knee Abduction ^  4  1.4.2^Knee Extension ^  5  1.4.3^Lower Limb Reaching ^  5  1.5 FEMALE RISK FACTORS ^  7  1.5.1^Anatomical Factors ^  7  1.5.2^Hormonal Factors ^  9  1.5.3^Neuromuscular Factors ^  9  1.6^INJURY PREVENTION PROGRAMS ^  12  1.7^SCREENING ^  14  1.7.1^Existing Screening Tools ^  15  1.7.2^Development of a Neuromuscular Screening Method ^ 20 1.8^PRINCIPLES OF MEASUREMENT ^  21  1.8.1^Reliability ^  21  1.8.2^Validity ^  23  ^  1.9 PURPOSE, OBJECTIVES AND HYPOTHESES ^  23  1.9.1^Purpose ^  23  1.9.2^Primary Objectives ^  24  1.9.3^Primary Hypotheses ^  24  2 METHODS ^  26  2.1^DESIGN ^  26  2.2^PARTICI PANTS ^  26  2.2.1^Recruitment ^  26  2.2.2^Inclusion/Exclusion Criteria ^  27  2.3 PROTOCOL ^  27  2.4 LABORATORY-BASED PROTOCOL ^  28  2.4.1^Set-Up ^  28  2.4.2^Performance Tasks ^  29  2.4.3^Instrumentation ^  37  2.4.4^Data Processing ^  38  2.4.5^Data Management ^  40  2.5 RATING PROTOCOL ^  42  2.5.1^Recruitment of Raters ^  42  2.5.2^Rater Training ^  42  2.5.3^Rating Procedures and Rules ^  43  2.5.4^Development of the Risk Screening Guidelines ^  44  2.5.5^Data Management ^  45  2.5.6^Observational Risk Screening Guidelines ^  46  2.6^STATISTICAL ANALYSIS ^  48  2.6.1^Test-retest Reliability ^  48  2.6.2^Rater Agreement ^  51  2.6.3^Validity ^  53  2.6.4^Relationship between Postural Measures and Knee Abduction ^ 55 2.6.5^Relationship between Gold Standard Variables ^  56  2.6.6^Sample Size Calculations ^  57  3 RESULTS ^ 3.1^PARTICIPANT CHARACTERISTICS ^  59 59 iv  ^  3.2^TEST-RETEST RELIABILITY ^  60  3.2.1^ICCs ^  60  3.2.2^Bland and Altman Methods ^  60  3.2.3^SEM ^  64  3.3 RATER AGREEMENT ^  64  3.3.1^I ntrarater Agreement ^  64  3.3.2^I nterrater Agreement ^  66  3.4 CUT OFF POINTS ^  66  3.5 VALIDITY OF OBSERVATIONAL RATINGS- METHOD 1 ^  68  3.5.1^Sensitivity ^  68  3.5.2^Specificity ^  70  3.5.3^Positive and Negative Predictive Values ^  71  3.6 VALIDITY OF OBSERVATIONAL RATINGS- METHOD 2 ^  72  3.7 RELATIONSHIP BETWEEN POSTURAL MEASURES AND KNEE ABDUCTION ^ 72 3.8 RELATIONSHIP BETWEEN GOLD STANDARD VARIABLES ^  4 DISCUSSION ^ 4.1^TEST-RETEST RELIABILITY ^ 4.1.1^Possible Factors Affecting Reliability ^ 4.2 AGREEMENT ^ 4.2.1^Possible Factors Affecting Agreement ^ 4.3^VALIDITY ^  73  74 74 75 76 78 81  4.3.1^Possible Factors Affecting Validity ^  84  4.4 SELECTION OF TASKS FOR RISK SCREENING ^  87  4.5^LIMITATIONS ^  88  5 CONCLUSION ^  91  6 BIBLIOGRAPHY ^  94  7 APPENDICES ^  109  7.1^APPENDIX I. CONSENT FORM ^  109  7.2^APPENDIX II. INFORMATION SHEET ^  114  7.3^APPENDIX III. LIMB LENGTH AND CIRCUMFERENCE MEASURES ^ 115 7.4 APPENDIX IV. POSTURAL MEASUREMENTS ^  116 V  7.5 APPENDIX V. IRED MARKER POSITIONS ^  117  7.6^APPENDIX VI. ETHICS CERTIFICATE OF APPROVAL ^  118  7.7^APPENDIX VII. RATING FORM ^  120  7.8 APPENDIX VIII. STANDARD ERROR OF MEASUREMENT ^  122  7.9^APPENDIX IX. ICC3, K FORMULAE ^  123  7.10 APPENDIX X. BLAND AND ALTMAN FORMULAE ^  124  7.11 APPENDIX XI. KAPPA SCALE OF AGREEMENT ^  125  7.12 APPENDIX XII. KAPPA:  2x2 CONTINGENCY TABLES AND CALCULATIONS ^ 126  7.13 APPENDIX XIII. KAPPA ASSUMPTIONS ^ 7.14 APPENDIX XIV. VALIDITY:  128  2x2 CONTINGENCY TABLES AND CALCULATIONS ^ 129  7.15 APPENDIX XV. MAGNITUDE OF CORRELATIONS ^  130  7.16 APPENDIX XVI. PREVALENCE OF PHYSIOTHERAPIST RATINGS ^ 131 7.17 APPENDIX XVII. PREVALENCE OF 'EXPERT' AND 'TRUE' RATINGS ^ 132 7.18 APPENDIX XVIII. ROC CURVE FOR DROP JUMP TASK ^  133  7.19 APPENDIX XIX. ROC CURVE FOR SIDE HOP TASK ^  134  7.20 APPENDIX XX. ROC CURVE FOR SIDE CUT TASK ^  135  7.21 APPENDIX XXI. CORRELATIONS BETWEEN KNEE ABDUCTION MOTION AND STATIC POSTURAL CHARACTERISTICS OF RIGHT LOWER LIMB ^  136  7.22 APPENDIX XXII FREQUENCY DISTRIBUTION- KNEE ABDUCTION MOTION ^ 137 7.23 APPENDIX XXIII FREQUENCY DISTRIBUTION- SIDE HOP REACHING ^ 138 7.24 APPENDIX XXIV FREQUENCY DISTRIBUTION- SIDE CUT REACHING ^ 139  vi  LIST OF TABLES  TABLE 1. SAMPLE SIZE REQUIREMENTS FOR DIFFERENT VALUES OF KAPPA ^ 58 TABLE 2. PARTICIPANT CHARACTERISTICS ^  59  TABLE 3. TEST-RETEST RELIABILITY OF MOTION ANALYSIS VARIABLES ^ 60 TABLE 4. INTRARATER AGREEMENT ^  65  TABLE 5. INTERRATER AGREEMENT ^  67  TABLE 6. VALIDITY: METHOD 1- 'TRUE RATINGS' AS GOLD STANDARD ^  69  TABLE 7. VALIDITY: METHOD 2- 'EXPERT RATINGS' AS GOLD STANDARD ^ 69  vii  LIST OF FIGURES  FIGURE 1. DYNAMIC KNEE ABDUCTION ANGLE ^  X  FIGURE 2. LOWER LIMB REACHING ^  XI  FIGURE 3. KNEE JOINT LIGAMENTS ^  3  FIGURE 4. LOWER LIMB REACHING ^  6  FIGURE 5. FEMALE INJURY RISK FACTORS ^  7  FIGURE 6. LOCATION OF IRED MARKERS ^  29  FIGURE 7. DROP JUMP TASK ^  30  FIGURE 8. DROP JUMP STARTING POSITION ^  31  FIGURE 9. SIDE HOP STARTING POSITION ^  32  FIGURE 10. SIDE HOP ^  33  FIGURE 11. SIDE CUT STARTING POSITION ^  35  FIGURE 12. SIDE CUT ^  35  FIGURE 13. CROSSCUT ^  36  FIGURE 14. FORMULA FOR NORMALISED REACHING RATIO ^  39  FIGURE 15. VERTICAL GROUND REACTION FORCE ACROSS STANCE PHASE ^ 40 FIGURE 16. SELECTION OF OUTLIERS FROM TIME NORMALISED OVERLAY PLOT (SIDE CUT) ^ 41 FIGURE 17. HIGH RISK AND LOW RISK DROP JUMP LANDINGS ^  46  FIGURE 18. HIGH RISK AND LOW RISK SIDE HOP LANDINGS ^  47  FIGURE 19. HIGH RISK AND LOW RISK SIDE CUT LANDINGS ^  48  FIGURE 20. MAXIMUM AND MINIMUM KNEE ABDUCTION ACROSS STANCE PHASE ^ 50 FIGURE 21. BLAND AND ALTMAN GRAPH FOR MAXIMUM KNEE ABDUCTION ANGLE ^ 62 FIGURE 22. BLAND AND ALTMAN GRAPH FOR KNEE ABDUCTION MOTION ^ 62 FIGURE 23. BLAND AND ALTMAN GRAPH FOR LOWER LIMB REACHING (SIDE HOP) ^ 63 FIGURE 24. BLAND AND ALTMAN GRAPH FOR LOWER LIMB REACHING (SIDE CUT) ^ 63  viii  LIST OF ABBREVIATIONS ACL:^Anterior cruciate ligament BMI:^Body mass index GRF:^Ground reaction force ICC:^Intraclass correlation coefficient IRED:^Infrared emitting diode LCL:^Lateral collateral ligament MCL:^Medial collateral ligament OA:^Osteoarthritis PCL:^Posterior cruciate ligament PPE:^Preparticipation Physical Examination PV+:^Positive predictive value PV-:^Negative predictive value ROC:^Receiver operating characteristic 3D:^Three-dimensional 2D:^Two-dimensional  ix  GLOSSARY ACL: anterior cruciate ligament; one of the main stabilising ligaments of the knee. Drop jump: performed by dropping to the floor from a 31cm box and then immediately  jumping vertically. Dynamic knee abduction angle: the angle formed between the tibia and femur in the  frontal plane as a result of hip adduction and internal rotation and knee flexion (FIGURE 1).  FIGURE 1. Dynamic knee abduction angle  High-risk movements: those movements such as changing direction and landing from  a jump that have most frequently been associated with ACL injuries. Joint moment: force causing rotation of a joint; 74 may be internal (e.g. internal  resistance provided by muscle) or external (e.g. external moment of gravity). Kinematics: study of motion of a body (e.g. joint angles). 74 Kinetics: study of the effects of forces on a body (e.g. joint moments).  74  Lower limb reaching: planting the foot without having the body's centre of mass  vertically aligned over its base of support (FIGURE 2).  FIGURE 2. Lower limb reaching  Medical problems precluding participant from testing: problems such as gastro-  intestinal illnesses, head colds, low blood pressure, chest infections, dehydration and menstrual pain that would present a barrier to participation in the two to three hour testing session. Neuromuscular control: interplay between the neural and muscular systems in  providing dynamic stability to a joint. Non-contact ACL injury: an anterior cruciate ligament injury not caused by a direct  blow to the involved lower extremity.4° Rater agreement: the likelihood that a rater will give the same rating to a subject as  another rater (interrater agreement) or that a rater will give the same rating to a subject across time points (intrarater agreement). Screening: identifying "those who are at sufficient risk of a specific disorder to justify a  subsequent diagnostic test or procedure, or in certain circumstances direct preventative action."58 (P  149)  xi  Serious injury: an injury requiring greater than or equal to 30 days off from the date of injury until when the player is able to return to full participation in team training and availability for match selection (adapted from Fuller et a127). Side cut: performed by striding forward and then quickly to the left or right. Side hop: performed by hopping laterally onto one leg and then back to the starting position. Significant injury: an injury requiring greater than or equal to 10 days off from the date of injury until when the player is able to return to full participation in team training and availability for match selection (adapted from Fuller et al27). Third-degree ligament injury: complete ligament rupture Validity: the likelihood that we are measuring what we purport to measure;8° in relation to screening, the likelihood of the rating being correct when compared with a 'gold standard' measure.  xii  ACKNOWLEDGMENTS Thank you to my supervisor, Dr. Bill Miller, for the faith you have shown in me over the last four years. You are an excellent teacher and mentor and I have learned a tremendous amount from you. Thank you to the members of my committee; to Dr. Donna Maclntyre, whose support, encouragement and guidance has been invaluable throughout this process and to Dr. Janice Eng, who is an inspiration to young investigators in this field. Thank you to Dr. Carolyn Emery, my external examiner, for giving up your valuable time to be part of this process. I owe an enormous debt of gratitude to Rick Celebrini for taking me on as your assistant and co-researcher and for your help in the conception of this study. You are a wealth of knowledge and a terrific person to work with - even during Sunday morning testing sessions. Your determination and tenacity in following through with your ideas and upholding high standards of research have inspired me to do the same throughout the rest of my career. Thank you to Tom Depew and JD Johnston for your technical support and programming wizardry. Thank you to Tom and also to Nicole Nadeau for frequently giving up your evenings and weekends to help with data collection. Thank you to Nadine Nembhard, Marilou Lamy and Dana Ranahan, PTs for your valuable contribution to this study. Your expertise and input were essential to the success of the study and were greatly appreciated. Thank you to the faculty and staff of the School of Rehabilitation Sciences and the other research graduate students, especially the members of the GF Strong research lab. I will miss you all. Thank you to all of the soccer players and parents for giving up your time to participate in this study; and to Bill Trenaman and Larry Moro from the Port Moody Soccer Club and Markus Reinkens from the BC Soccer Association for your help with recruitment. Finally, I want to say a big thank you to all of my family and friends back in Australia for your continuous support, and to Paul, for coming to Vancouver to be with me throughout this process and for helping to keep it all in perspective.  I INTRODUCTION 1.1 ACL Injuries — The Scope of the Problem Of all athletic knee injuries an anterior cruciate ligament (ACL) rupture is the most devastating; resulting in the greatest time lost from sport,15 greatest financial costs13 and most significant long term problems.92• 107 The ACL plays a vital role in the normal function and stability of the knee. Therefore, individuals wishing to return to sport after injury are encouraged to consider surgery to repair the ACL. Following this, up to 12 months of post-operative rehabilitation are required to recover pre-injury levels of agility and function. During the post-operative period, athletes will experience significant deconditioning and a loss of sport-specific skills. Even with a successful rehabilitative outcome, on returning to sport, an athlete with a previous ACL injury has a significantly greater risk of experiencing another ACL injury. 79, 108, 111  Those who opt out of surgery are often forced to reduce their level of physical activity and their involvement in sport." This can have serious implications later in life, with reduced physical activity being associated with a higher incidence of obesity, morbidity and mortality.^• 81 Of most concern is that an ACL injury will triple the risk of osteoarthritis (OA) developing by middle age. 89 This disease can have a significant impact on quality of life by further limiting functional and leisure activities.14•  35  It is  clearly of utmost importance that efforts be directed towards front line prevention of these serious injuries.  1  1.1.1 Non Contact ACL Injuries Only 30% of athletic ACL injuries result from some form of contact during play.34 The remaining 70%, termed non-contact injuries, are most commonly caused by decelerating before a change of direction or landing from a jump on one leg.  5 78 '  Thus,  sports such as soccer and basketball which involve frequent jumping, landing and cutting have a high incidence of this form of ACL injury.  '  13 84  ACL injuries have become  the most prevalent third degree ligament injury at the knee,49 and although world-wide incidence estimates are not currently available, Griffin et al estimate that in the US, the incidence of ACL injury is as high as 100 000 per year.34 In 1999, the estimated surgical and rehabilitation costs per injury totalled approximately $17 000 USD (F. R. Noyes, unpublished data, 1999). Today, this would equate to almost two billion dollars (US) of health care spending annually.  1.1.2 ACL Injuries and the Female Athlete Most alarming is the high incidence of ACL injury in female athletes. Women have a four to six times higher risk of non-contact ACL injury than men.  2, 22, 33, 84  Soccer is a  particularly high risk sport for this population, with women being more likely to injure their knees while playing soccer than in any other team sport.13 Of all soccer-related knee injuries, cruciate ligament injuries are among the most common,13 and with female participation in soccer continuing to increase every year, the incidence of ACL injury in this population is on the rise. 15, 24, 103  2  1.2 ACL Anatomy The ACL is one of four major ligaments of the knee; the other three being the posterior cruciate ligament (PCL) and the medial and lateral collateral ligaments (MCL/LCL) (FIGURE 3). All four ligaments provide passive stability to the knee joint, with the ACL being primarily responsible for resisting anterior translation and internal rotation of the tibia on the femur. The ACL arises from the posterior part of the lateral wall of the intercondylar notch of the femur and inserts onto the anterior intercondylar area. It is composed of dense irregular connective tissue and has a relatively poor blood supply.  71  FIGURE 3. Knee joint ligaments  @ Silbey MB, Fu FH. Knee Injuries. In: Fu, F.H. and Stone, D.A. (Eds.), Sports Injuries. Mechanisms, Prevention, Treatment. Baltimore: Williams and Wilkins; 1994, p. 949-976, by permission.  3  1.3 ACL Injury Mechanisms Non-contact ACL injuries most commonly result from decelerating before a change in direction or landing from a jump on one leg.8'  75  The mechanism of ACL injury during  these movements may include a combination of the following components: 1) dynamic knee abduction  ;40, 62, 68, 93  2) a knee that is close to full extension;  62' 93  and  3) lower limb reaching (a centre of mass that is not aligned vertically over the planted foot).98'  115  Numerous biomechanical studies have shown that even in a controlled laboratory setting, women have a tendency to perform jump landings and side cuts with a less flexed and more abducted knee than men.24-26,  41, 51, 66, 70, 91, 93, 113, 114  A gender  difference has not yet been explored in relation to lower limb reaching .  1.4 Effect of these mechanisms on the ACL 1.4.1 Dynamic Knee Abduction To abduct the knee, the femur must adduct and internally rotate which causes a relative external rotation of the tibia. As the femur internally rotates, the lateral wall of the femoral notch may impinge and then sever the ACL.28 The high quadriceps forces produced on landing or decelerating apply an added anterior translatory pull to the ACL which further increases the risk of rupture.1 Supporting the findings of biomechanical and anatomical studies, a recent prospective study of adolescent female athletes found 4  that landing from a drop jump with an increased knee abduction angle and external abduction moment can together predict a future ACL injury.41  1.4.2 Knee Extension When the knee is positioned near full extension, the posterior fibres of the ACL are most taut.96 These fibres provide the primary resistance to hyperextension and are most susceptible to injury while the knee is extended.64 When abduction loads are then exerted on the knee, the ACL is put under further strain.96  1.4.3 Lower Limb Reaching During single limb landing tasks, if the lower limb is positioned neutrally with the body's centre of mass vertically aligned over its base of support, any excess forces will be absorbed and dampened via hip and knee flexion and ankle dorsi flexion. These sagittal plane movements cannot injure the ACL.66 However, if the limb is loaded while in an abducted position, the centre of mass will be outside of the base of support and potentially harmful forces can be transmitted through the kinetic chain.  In a recent study by Sigward et al, lower limb mechanics during a side cut were examined in female soccer players aged 14 to 18.96 Subjects with high knee abduction moments had greater hip abduction at initial contact than those with 'normal' knee abduction moments. The authors hypothesised that hip abduction moves the foot's centre of pressure lateral to the tibia's centre of mass. This creates a larger moment arm for the vertical intersegmental component of the ground reaction force (GRF) at the 5  distal tibia and thereby exerts a larger abduction moment on the knee (FIGURE 4). This tendency to plant the foot with the centre of mass outside the base of support is commonly observed when an athlete is decelerating prior to a change in direction or when a soccer player is trying to kick a ball. It is referred to as "reaching" (FIGURE 4) (R. Celebrini, 2007, unpublished thesis in progress).  Vertical GRF  FIGURE 4. Lower limb reaching  6  1.5 Female Risk Factors There are numerous theories as to why these high risk movement patterns may be more common in women and why they are more likely to result in an ACL rupture in this population. The causative factors have been divided into three main categories, 1) anatomical, 2) hormonal and 3) neuromuscular (FIGURE  5). 42  Anatomical factors  Hormonal^Neuromuscular factors^factors  FIGURE 5. Female injury risk factors  1.5.1 Anatomical Factors During puberty, girls undergo many anatomical changes which include a widening of the pelvis, an increased Q (quadriceps) angle, increased knee hyperextension, increased external tibial torsion and more forefoot pronation. 6 Many of these anatomical characteristics contribute to faulty alignment positions that are similar to reported ACL injury mechanisms. 6 When individuals with poor alignment adopt high risk dynamic positions, they are thought to be more at risk of injury because the ligaments are already preloaded and near their limits of tensile strength. 6 In a study by Loudon, knee hyperextension combined with an excessive navicular drop and excessive subtalar joint  7  pronation were found to be significant discriminators between ACL-injured and noninjured individuals.60 In contrast, Gray et al performed a retrospective study of female basketball players with injured ACLs and found there was no relationship between knee alignment and knee injury.33  These authors also investigated the relationship between Q angle and knee injury. The Q angle is defined as the angle between the line connecting the anterior superior iliac spine and the midpoint of the patella, and the line connecting the tibial tubercle and the patella.17 Wider hips contribute to a greater Q angle in women and while it has been linked with a greater incidence of other knee injuries, such as patellofemoral pain syndrome,19'  87  there is no conclusive evidence linking the size of the Q angle with ACL  60 94  •^33 injury. ' ' In light of conflicting evidence, the exact contribution of anatomical factors  to ACL injury remains uncertain.  Beyond lower extremity structure and alignment, general physical factors have also been linked with greater injury risk. There is some evidence that women with a greater body mass index (BMI) are more at risk of ACL injury. 106 It has been suggested that these individuals may have under-developed neuromuscular control and coordination patterns resulting from a history of reduced physical activity. The same study also found a link between greater whole body joint laxity with an increased risk of ACL injury. 106 However, several previous studies failed to find a similar relationship.  30 3 2, 50  '  8  1.5.2 Hormonal Factors The exact link between female hormonal factors and ACL injury is not yet fully understood. An increased risk of ACL injury has been found by certain authors during the follicular and ovulatory phases of the menstrual cycle.99'  199  Oestrogen and  progesterone receptors have been found on the ACL and it is thought that fluctuations in these hormones might influence the tensile strength and proprioceptive function of the ACL.17'  59  It is unlikely that hormonal factors alone could cause an ACL injury. However,  combined with anatomical and neuromuscular factors, the influence of hormones could certainly increase the risk.21'  24  1.5.3 Neuromuscular Factors Unlike anatomical and hormonal factors, neuromuscular factors appear to be modifiable and therefore have become the focal point of the recent push for ACL injury prevention.  36, 39, 63, 73, 82, 100  Neuromuscular factors relate to the interplay between the  neural and muscular systems in providing dynamic stability to a joint. Neuromuscular risk factors in women can be further categorised into:  1) lower limb imbalances; and 2) proximal control deficits  Lower limb imbalances include limb dominance and quadriceps dominance. Limb dominance refers to having significant side-to-side differences in lower limb strength and neuromuscular control. Limb dominance is more marked in women and appears to 9  develop after the onset of puberty.40'  53  This mismatch puts both limbs at higher risk of  injury, with the dominant limb being over-utilised and therefore put under more strain and the non-dominant limb being under-utilised and thereby made weaker and less resilient.40'  41  Quadriceps dominance refers to a mismatch in strength and timing between the quadriceps and hamstrings. An unopposed quadriceps contraction causes an anterior translation of the tibia on the femur, which can strain the ACL.1 This reliance on the quadriceps over the hamstrings for knee joint stability is commonly seen in female athletes." The imbalance may partly result from the female tendency to land with the knee less flexed. When the knee is close to full extension the hamstrings may be inhibited, allowing the quads to dominate.2  Proximal control deficits are thought to be the primary cause of the high risk movement patterns commonly seen in women.  115, 116  Dynamic knee abduction, which  should not be confused with postural genu valgum, results from femoral adduction and internal rotation during tasks involving hip and knee flexion.45 This unwanted femoral adduction and internal rotation are thought to result from a lack of strength and control in the musculature of the hip.  Claiborne et al demonstrated that the peak torque of the hip abductors was a significant predictor of frontal plane knee motion during a single leg squat, suggesting that the hip muscles play a major role in controlling knee abduction motion. They also found that  10  male subjects exhibited significantly greater absolute peak torque for all major hip muscle groups compared to females. 1° Jacobs et al studied jump landings in male and female participants and found stronger negative correlations between hip abductor strength and knee abduction motion in females than males. 46 This suggests that females, with their tendency towards increased postural and dynamic knee abduction, may be more reliant on hip abductor strength for providing neuromuscular control to the knee.  In contrast with these findings, a recent study failed to find a correlation between hip muscle strength and the amount of frontal plane knee movement during a forward lunge. 104 The authors suggested that the degree of knee abduction during dynamic tasks is not purely dependent on the strength of the hip musculature but that other factors such as proprioception and core stability might be more important. This idea was echoed by Zazulak et al, who stated that proprioceptive deficits in the body's core may contribute to decreased active neuromuscular control of the lower limb and the associated knee abduction. 116  Reduced neuromuscular control may also be manifested as a more extended knee during a jump landing or side cut. Rather than utilising the hip and knee musculature eccentrically to control hip and knee flexion, female athletes appear to rely more on the bony and ligamentous restraints of the extended knee to brake the movement. 21 ' 24  11  As an ACL risk factor, lower limb reaching has received little attention compared to the phenomenon of dynamic knee abduction. There have been no laboratory studies which have measured the degree of lower extremity reaching in the general population. Therefore, it is unknown whether this high risk movement pattern is more common to women. It is also unknown why some athletes display these tendencies and others do not. One theory is that in some individuals there may be deficiencies in the proprioceptive system that provides feedback to the central nervous system about the position of the joints and the body as a whole.115 If the position of the body is represented inaccurately, insufficient or inappropriate descending movement commands may result.115 This may affect control of the trunk and lower limb and may lead to knee injury.  There is also support for the idea of a 'neuromuscular spurt' occurring to differing degrees during puberty. This neuromuscular spurt is described as an increase in neuromuscular control and muscle mass which leads to improved coordination, faster reflexes and more efficient motor patterns.37' 4°'  85  Women do not appear to experience a  neuromuscular spurt to the same extent as men. Therefore, post-pubescent men may be better at controlling their taller, heavier frames and at protecting themselves when playing high level sports!' 4°  1.6 Injury Prevention Programs Based on their findings of increased hip abduction (reaching') in female soccer players with greater knee abduction moments, Sigward et al recommended that ACL prevention 12  programs be designed to encourage loading of the lower extremity in the sagittal plane with neutral rotation.96 They suggested that correct body alignment with a more vertical tibia would decrease frontal plane loading and shear forces and reduce the risk of injury. Several studies have demonstrated improved neuromuscular characteristics and reduced ACL injury in individuals who have undergone training programs aimed at improving body alignment and biomechanics. 2,36,39,63,73,82,100  In a recent pilot study, female soccer players underwent a four week training program aimed at reducing their tendency to reach (R. Celebrini, unpublished results, 2007). The participants were taught to "lead with the centre" and produce a stronger push off the back leg while performing soccer-specific drills. Early observations of results appear to indicate that, in those players who successfully learned the technique, potentially harmful knee abduction moments were reduced when performing side cut and side hop movements. The results also suggest that when an individual was able to land or plant with the centre of mass over the base of support, the knee flexion angle was increased.  Arendt and Dick reported on an unpublished study in which athletes were taught to perform a side cut as a three-step stop with the knees flexed.2 When this technique was taught to two division-one female basketball teams, the incidence of ACL injuries was reduced by 89% over several years. The authors hypothesised that the technique of keeping the knee flexed and the feet more under the hips when cutting and turning, helped the athletes avoid terminal extension where the ACL is in a more stretched position. They also speculated that a flexed knee position facilitates contraction of the hamstrings which may help to protect the ACL against rotatory and translatory forces.72 13  A recent systematic review identified six randomized controlled trials which investigated the effect of neuromuscular training on ACL injury incidence in female athletes.38 Through the use of plyometrics, balance and strengthening exercises, three of the six studies reported improvements in neuromuscular risk factors and significant reductions in the incidence of ACL injury. These three studies differed from the others in that they used analysis of movement biomechanics and provided regular feedback to athletes regarding proper body position and technique.38  1.7 Screening According to Lang and Secic, the aim of screening is:  "to identify those who are at sufficient risk of a specific disorder to justify a subsequent diagnostic test or procedure, or in certain circumstances direct preventative action.  768(p149)  In order to reduce ACL injury, several authors have recommended that female athletes be screened for the presence of neuromuscular risk factors.8'  37' 41' 69  Those displaying  high-risk neuromuscular characteristics, such as increased knee abduction angles or lower limb reaching, could then be targeted for injury prevention initiatives. The efficiency and cost-effectiveness of intervention programs would then be optimized by ensuring that they were delivered to those most in need.24  In a study by Myer et al female athletes were put into high risk and low risk groups based on the presence of high risk knee kinematics and kinetics.72 They found that high 14  risk female athletes decreased the magnitude of their knee abduction moments following neuromuscular training to a greater degree than low risk athletes. The authors recommended targeting individuals with high risk characteristics in order to benefit those most in need of training.  1.7.1 Existing Screening Tools In amateur and youth sports, where there is a high rate of injury, screening is uncommon.41 This may be due in part to a lack of efficient and effective screening tools available to coaches and clinicians.  The most well known screening tool in sport is the Preparticipation Physical Evaluation (PPE) which is mandated for use in all US high school and college sports programs.8 In recent years, the musculoskeletal component of the PPE has come under strong criticism for having little ability to detect existing injuries and no proven predictive value.8'  29,31  It consists mainly of static measures of joint stability and is completely  devoid of any items that provide qualitative feedback on neuromuscular contro1.24 The PPE was designed as a broad-based tool capable of screening for a wide range of different types of injuries. By developing a tool that would 'cover all bases' the creators of the PPE sacrificed its sensitivity.  To increase sensitivity, it has been suggested that screening tools focus on the risk factors of specific injuries in specific populations.9° Neuromuscular risk factors have become the focus of the push for ACL injury prevention. Therefore, it is vital that a 15  reliable, accurate and practical method of screening athletes for high risk neuromuscular characteristics be made available to clinicians and coaches.  Recently, researchers from Cincinnati, USA set out to examine whether certain neuromuscular risk factors could predict ACL injury.  41  In a prospective cohort study,  they followed 205 adolescent female soccer, basketball and volleyball players for 13 months. At baseline they used three dimensional (3D) motion analysis and force plates to collect kinematic and kinetic data on subjects performing a drop jump landing. Players who went on to experience an ACL injury during the 13-month surveillance period displayed a mean peak knee abduction angle of 9° during the drop-jump landing. Those who did not experience an ACL injury displayed a significantly lower mean peak abduction angle of 1.40. Increased knee abduction angles combined with increased knee abduction moments were reported to be predictive of ACL injury in this population and it was recommended that athletes be screened for these characteristics prior to sports participation.  Several researchers have since explored the possibility of screening athletes for high risk kinematics and kinetics without the need for sophisticated motion analysis techniques.9'  37' 69' 77  Based on the theory that knee abduction moments are correlated  with ground reaction forces, Hewett et al used a portable force plate to screen for ACL injury risk in 275 adolescent athletes.37 They found that female athletes were unable to properly attenuate landing forces regardless of their stage of pubertal development. In contrast, the landing characteristics of male athletes improved with maturation. The  16  authors recommended that portable force plates be used as part of a preseason screening in order to detect high-risk landing profiles in young athletes.  Noyes et at used simple video cameras to record participants performing a drop jump landing!' From the video footage, the researchers measured the separation distance (in cm) between participants' knees, normalised to their hip width. They deemed that 560% normalised knee separation distance (knee separation/hip width) was indicative of a significant degree of lower limb valgus (knee abduction). This method had high testretest reliability (ICC5>0.94). However, the researchers did not attempt to validate this technique though comparison with a gold standard technique, such as 3D motion analysis.  McLean et al investigated the interchangeability of two-dimensional (2D) and 3D motion analysis methods.69 Participants performed tasks such as a side-step cut and a side jump to enable investigation of the relationship between knee abduction angles measured using the conventional 3D approach and those measured with simpler 2D techniques. The 2D video analysis technique produced consistent data that correlated moderately well with 3D motion analysis data.  While these three screening methods are certainly easier to administer than 3D motion analysis, their application remains limited. To promote screening in the field or clinical setting, screening tools need to be quick and easy to use and not require sophisticated equipment. In the study by Hewett et al, the use of a portable force plate was  17  advocated.37 However, the high cost of purchasing such a device would be prohibitive to most clinicians and coaches. Also, for all three methods outlined above, lengthy posthoc data analysis was required, meaning that these screening methods would only be feasible to those with access to computerised data analysis programs and technical assistance.  Observational risk screening is a far more time and cost-effective option, enabling coaches to give frequent and immediate feedback to their players during the course of normal training and providing clinicians with a quick and simple assessment tool. To our knowledge, only two studies have investigated using observational screening methods to detect high risk neuromuscular characteristics.  Chmielewski et al studied the reliability of using observational rating guidelines to assess the degree of frontal plane trunk and lower limb motion exhibited during a unilateral squat and a lateral step-down task.9 The screening tasks chosen in this study were performed in a slow and constrained manner in a controlled environment which limits the relevance of this screening method to athletes. Screening tasks must be dynamic enough to challenge athletes and should attempt to replicate the high-risk manoeuvres of their particular sport.52  In an attempt to better replicate real, high risk conditions, Krosshaug et al investigated the accuracy and precision of estimating lower limb kinematics from video sequences of situations resembling those typically leading to ACL injuries.55 Raters were asked to  18  estimate hip, knee and ankle joint angles at predetermined points on the video. The researchers also examined whether raters could visually detect a change in joint angle between two time points by asking them to classify the change in knee angle as valgus, neutral or varus. In this study, raters were permitted to rewind, pause and view the footage in slow motion. While these leniencies may have helped to improve reliability and validity of ratings, they prevented extrapolation of the results to real clinical screening conditions. If screening methods are to be used by clinicians and coaches in the field it is vital that the methods be tested under conditions that are as close to real life as possible.  In summary, for a screening method to be recommended for use in ACL risk screening it must: •  be feasible for use in the field or clinical setting, not requiring expensive equipment,  •  be quick to administer,  •  include challenging, dynamic tasks and  •  have acceptable reliability and validity  This study aims to develop a screening method meeting all of these criteria.  19  1.7.2 Development of a Neuromuscular Screening Method So far, any attempts at clinical ACL risk screening have focused on only one risk factorincreased knee abduction angles. However, knee abduction is not involved in all ACL injuries and not all individuals at risk of ACL injury exhibit this characteristic.56 To ensure risk screening encompasses a wide range of risk factors, it is important that other factors are explored. These individual components could eventually contribute to the creation of a more comprehensive risk screening package. In this study, two separate risk factors will be explored.  1. Dynamic Knee Abduction Observational risk screening guidelines will enable raters to classify participants as high risk or low risk based on the degree of dynamic knee abduction exhibited on a drop jump landing. To perform a drop jump, an individual jumps down onto the ground from a box and then performs a maximal vertical jump.  2. Lower Limb Reaching The second component of the observational risk screening guidelines will allow raters to classify participants as high risk or low risk based on the degree of lower limb reaching exhibited during a side hop and a side cut. To perform a side hop, an individual hops to the side, lands on one leg and then hops back to the starting position. The side cut consists of a stride forward followed by a stride to the left or right to kick a soccer ball.  20  1.8 Principles of Measurement Before a new screening method can be recommended for use in practice and research, its measurement properties must be critically evaluated. This evaluation process includes investigating whether acceptable standards of reliability and validity have been attained. In this study three forms of reliability and one form of validity will be investigated.  1.8.1 Reliability Portney and Watkins state:  "Reliability is fundamental to all aspects of clinical research, because without it we cannot have confidence in the data we collect, nor can we draw rational conclusions from the data."33  (P 61)  Reliability reflects the extent to which measures are free from error.8° Error in measurement (E) is defined as the amount that an individual's observed score (X) differs from their 'true score' (T); that is, the average score they would get from infinite testing (see equation 1).80  X=T+E  Equation 1: Linear model of true score theory 80  21  Sources of measurement error In performance based measures, there are multiple potential sources of error. Firstly, an individual's performance may change from one day to the next or even from one instant to the next. This participant-related error can result from fatigue, inattention, learning or environmental distractions. Also if the testing protocol is not well defined, error can be introduced by providing inconsistent or unclear instructions to participants or by not controlling the environment. A consistent location should be used for repeated testing and the participant should be consistently attired. The use of technical equipment for data collection also introduces a potential source of error. With 3D motion analysis, there may be excessive marker motion, markers may be inconsistently positioned, sampling frequencies may be inappropriate or computer programmes may malfunction. To evaluate the influence of all of these sources of error on individuals' motion analysis scores, test-retest reliability will be reported.  When raters are evaluating an individual's performance an additional source of error is introduced. Rater-related error may be caused by perceptual differences between raters as a result of differing opinions, level of training or level of experience. These differences between raters will be reflected by interrater reliability. Individual raters may also be inconsistent over time as a result of fatigue, inattention, learning or environmental distractions. This will be reflected by intrarater reliability. Studies of intra or inter-rater reliability are more accurately described as investigations of agreement among or between raters.98 Therefore from this point on, rater reliability will be referred to as rater agreement.  22  1.8.2 Validity Validity in measurement refers to the accuracy of our inferences.80 In other words, that we are measuring what we purport to measure. The validity of a screening tool can be determined by comparing the test result with known diagnostic findings obtained from a 'gold standard' outcome measure. In this study, 3D motion analysis of knee abduction and lower limb reaching will act as the 'gold standard'. Physiotherapists' ratings will be compared with results from 3D motion analysis to obtain sensitivity and specificity values.  Sensitivity reflects the proportion of people with the disorder who have a positive test result.12 Specificity reflects the proportion of people without the disorder who have a negative test result. In this study, the 'disorder' of interest is a high degree of knee abduction or lower limb reaching (measured with 3D motion analysis). The 'test result' of interest will be an observational rating of 'high-risk' or 'low-risk'.  1.9 Purpose, Objectives and Hypotheses 1.9.1 Purpose The purpose of this study is to examine the agreement and validity of using observational risk screening guidelines to evaluate ACL injury risk factors.  23  1.9.2 Primary Objectives 1. To devise observational risk screening guidelines for the evaluation of two neuromuscular characteristics associated with ACL injury: i. dynamic knee abduction; and ii. lower limb reaching  2. To examine test-retest reliability of measuring dynamic knee abduction and lower limb reaching using 3D motion analysis  3. To examine the intra-rater and inter-rater agreement of observational ratings  4. To examine the validity of using observational risk screening as an alternative to 3D motion analysis  1.9.3 Primary Hypotheses 1. Test-Retest Reliability: 3D motion analysis of knee abduction and lower limb reaching will be measured with acceptable reliability according to Bland and Altman Methods of Agreement (minimal bias; clinically acceptable limits of agreement) and intraclass correlation coefficients (ICCs ^0.75, significant at the .05 level).83  24  2. Intrarater Agreement: The observational ratings of each physiotherapist between time 1 and time 2 will demonstrate at least a 'substantial' level of agreement (kappa 0.61, significantly greater than zero at the .05 level (1-tailed)).  3. Interrater Agreement: The observational ratings between physiotherapists will demonstrate at least a 'substantial' level of agreement (multirater and standard kappa 0.61, significantly greater than zero at the .05 level (1-tailed))  4. Validity: Observational risk screening will detect individuals with high-risk knee abduction angles and lower limb reaching with a high degree of sensitivity ( ^0.80) and moderate specificity ( ^0.50).  25  2 METHODS 2.1 Design This is a methodological study consisting of four main parts: 1. Development of observational risk screening guidelines for the evaluation of high  risk neuromuscular characteristics of the lower limb  2. Test-retest reliability of 3D motion analysis methods  3. Agreement of observational ratings, including: -intrarater agreement of observational ratings -interrater agreement of observational ratings  4. Validity of observational risk screening  2.2 Participants 2.2.1 Recruitment  Between January and June 2007, 40 female participants were recruited using a convenience sampling approach from soccer teams in British Columbia, Canada. Coaches of national-level U14-16 teams and elite club (gold-level) U14-16 teams were  26  contacted by the investigator to request permission to speak to the players during one of their training sessions. At this session, the investigator explained the study and distributed information and consent forms for the players and their parents/guardians to read (APPENDIX l). Following this, approximately 70 players were contacted by email and/or by phone to answer any study-related questions and assess suitability for inclusion.  2.2.2 Inclusion/Exclusion Criteria Participants were included if they 1) were aged between 13 and 17 years and 2) played soccer at a competitive level. They were excluded from the study if they had 1) a history of a 'serious' back or lower limb injury in the past, 2) any 'significant' back or lower limb injury in the six weeks prior to testing, or 3) any consistent 'medical problems' preventing participation in testing (see definitions, pp.xi-xii).  2.3 Protocol This study used two protocols, one for laboratory-based procedures and one for rating procedures.  27  2.4 Laboratory-Based Protocol 2.4.1 Set-Up After obtaining written and informed consent, participants were scheduled for a data collection session. All data collection occurred at the GF Strong Rehabilitation Research Laboratory, located in Vancouver, Canada. Upon arrival at the laboratory, participants were asked to fill out an information sheet with their demographic details, injury history and limb dominance (APPENDIX II). Limb dominance was determined by asking the participant with which leg they would prefer to kick a ball. Next, participants changed into a standardised pair of tight-fitting Lycra shorts and their own low-cut running shoes.  For the purposes of kinematic analysis, height, weight, limb segment length and limb circumference were measured. Height was measured in centimetres (cm) and weight was measured in kilograms (kg) using a height rod and mechanical balance scale (Health-O-Meter, Continental Scale Corporation, Bridgeview, Ill., USA). The length and circumference of the leg and foot were measured in cm using a cloth tape-measure at points defined by Yeadon and Morlock (APPENDIX 111).  112  A postural assessment was  performed to permit later correlational analyses of the relationship between postural characteristics and neuromuscular control. Bilateral rear foot eversion, inversion, knee extension and knee flexion were measured in degrees with a manual goniometer using standardised methods (APPENDIX IV).75 Navicular drop, knee valgus and knee varus were measured in centimetres with a plastic ruler using methods described by Magee (APPENDIX IV).61  28  Next, participants had 12 infrared emitting diodes (IREDs) placed on the pelvis, thigh, leg and foot of the dominant lower limb as described by Jian et al. and Eng and Winter (FIGURE 6, APPENDIX V).20'  48  These were used to track the participants' lower limb  motion to generate 3D kinematics and ultimately enable validation of observational ratings. Upon completion of set-up and measurement of anthropometric data, participants performed nine consecutive trials of each of three performance tasks — a drop jump, a side hop and an unanticipated side cut.  FIGURE 6. Location of IRED markers  2.4.2 Performance Tasks Drop Jump The drop jump task was taught to participants according to the protocol used in the study by Hewett et al that linked increased knee abduction angles with future ACL 29  •^41  injury. The task is straightforward and easily replicated and has been used by several other researchers in this field to elicit high risk knee abduction angles in young athletes.24 40, "  To perform a drop jump the participant jumps down onto the ground from a 31cm box and then immediately performs a maximum vertical jump (FIGURE 7). Unlike in Hewett et al's protocol, participants were instructed to keep their arms abducted to the 'stop position' throughout the jump to reduce momentum due to arm swing (FIGURE 8). Standardised instructions were given to each participant as follows:  "On the command go, you will jump down onto the ground from the box, ensuring your right foot lands on the force plate, and then jump up in the air as high as you can."  FIGURE 7. Drop jump task  30  FIGURE 8. Drop jump starting position  The jumping box was positioned 15cm behind a force plate embedded in the ground along the plane of progression (Bertec, Columbus, Ohio, USA). To ensure that only the right foot came into contact with the force plate on landing, the box was positioned so that its right edge was aligned with the centre of the force plate (FIGURE 8).  Side Hop The side hop task was taught to participants according to a novel protocol adapted from Noyes et al's Cross-over Hop Test. 76 In their test, the athlete hops 3 times side-to-side on one limb over a 15 cm wide centre strip. The hopping distance is measured and compared between limbs. We felt that this task provided a good combination of the two most common ACL injury mechanisms, namely, landing on one leg and changing direction.  31  We adapted the protocol to include only one hop rather than three, and since we were assessing landing kinematics rather than hopping distance, we normalised the starting position to the participant's height. The starting position was with the left foot on the outer edge of a line marked at a distance of 70% of the participant's height directly to the left of the centre of the force plate. The right foot was positioned slightly closer than hip-width apart from the left foot (FIGURE 9). Our choice of starting position was based on pilot testing with six different participants of varying ages and skill levels and represented a distance that was challenging but still within reach. Standardised instructions were given to each participant as follows:  "On the command 'go' and as quickly as you can, you will hop onto your right foot, landing in the centre of the force plate. You will then hop back onto your left foot, landing in the starting position" (FIGURE 10).  .00. MN  •■■■  ■ft,  .. .... .....  Force plate 70% height  FIGURE 9. Side hop starting position  32  FIGURE 10. Side hop  Unanticipated Side Cut The unanticipated side cut protocol was adapted from the task used in Sigward and Power's study. 96 In their study, female participants were found to perform this task with varying degrees of hip abduction ('reaching'). We therefore felt that the side cut would be a suitable screening task for identifying individuals who 'reach'. The side cut also replicates the change of direction commonly implicated in ACL injuries.  To further simulate game-like conditions, the side-cut included an unanticipated component whereby the participant was instructed to cut left or right depending on the last-minute directional cue received. Introducing a reactive component to testing has been shown to reveal high-risk strategies to a greater degree. 67 ' 93 Having to perform a sudden change of direction, athletes are less able to concentrate on their technique and may unwittingly expose the unsafe characteristics that increase their risk during play.  Owing to equipment limitations we were unable to include the five metre run up advocated by Sigward and Powers. 96 Instead, the side cut was preceded by a single step forward onto the force plate. Again, the task dimensions were chosen based on 33  pilot testing and our desire to create a feasible, yet challenging manoeuvre. The participant started with her feet in a staggered stance position with the right foot 50cm in front of the left (FIGURE 11). The toes of the left (take-off) foot were positioned 50 cm to the left and 83, 93 or 103 cm behind the centre of the landing force plate, depending on whether the participant's height fell between150-160 cm, 160-170 cm or 170-180 cm, respectively. Standardised instructions were given to each participant as follows:  "On the command 'go', and as quickly as you can, you will step backwards with your right foot then stride forward so that your entire right foot lands on the force plate. You will then use your left foot to kick the ball into the net. The arrow will tell you whether to kick the ball on the right or the left."  As the participant stepped back with her right foot a laser beam was broken, triggering a direction arrow informing her whether to kick left or right. If the participant received a left arrow she had to perform a side cut to kick the ball on the left and if she received a right arrow she had to perform a crosscut, across her body to kick the ball on the right (FIGURE 12 and 13).Test trials were randomised to either direction, ensuring an unanticipated component to the task and the appearance of the arrow was timed so as to force the participant to react quickly. The arrow appeared on a monitor positioned at approximately chest height, 275 cm forwards of the centre of the force plate. The left sided ball was positioned at a distance 80% of the participant's height away from the centre of the force plate and at an angle 55° anticlockwise to the plane of progression. This angle has been shown to be within the range of those cutting angles occurring most frequently during game situations.68 The right sided ball was positioned 80 cm 34  FIGURE 11. Side cut starting position  FIGURE 12. Side cut 35  FIGURE 13. Crosscut  from the centre of the force plate and at an angle of 150 anticlockwise to the plane of progression. Only the side cutting (left arrow) and not the crosscutting (right arrow) kicks were used for analysis as the former are more commonly involved in ACL injury mechanisms.  For each task, participants were allowed three practice trials prior to performing the nine test trials. The practice trials allowed an adaptation period to minimise variability due to learning effects. To prevent fatigue, participants adhered to a work/rest ratio of 1:5. With each trial taking a maximum of two seconds to complete, participants were given rest periods of ten seconds between each trial. This ratio has been successfully implemented in previous studies of this type.45 Participants were asked to repeat the trial if they lost their balance, if their right foot did not land on the force plate or if they did not perform the task correctly (e.g. allowing arms to move from the stop position during the drop jump or hesitating to receive the directional cue before kicking).  On average, the testing procedures took two hours to complete. To examine the testretest reliability of these procedures, ten participants were tested twice, one week apart. 36  The other thirty participants were tested only once. The protocol for this study was approved by the University of British Columbia Clinical Research Ethics Board and the rights of all participants were protected (see APPENDIX VI for ethics certificate of approval).  2.4.3 Instrumentation Optotrak© and Force Plate 3D kinematics of the lower limb were measured using an Optotrak© motion analysis system consisting of two cameras containing three sensors each (Optotrak 3020, NDI, Waterloo, Ontario, Canada). Data was sampled at 120Hz. Standardised joint coordinate systems for each segment were defined using digitised landmarks. Three non-collinear markers were used to track each body segment to generate a 3D model of the lower body. Kinematic data was collected with the participant first standing in a static neutral position with the feet hip-width apart to define zero. The force plate was synchronised with the Optotrak to delineate stance phase kinematics. The stance phase was defined as the period from foot contact to toe off. Force plate data was sampled at 600 Hz.  Digital video A digital camcorder (Canon ZR800A, Canon, Lake Success, NY, USA) was used to capture video footage for later viewing by raters. Video was recorded at 60 Hz. The camera was set up on a table mounted tripod 150 cm off the ground and 300 cm forwards of the centre of the force plate aligned perpendicular to the plane of  37  progression. The camera was positioned so as not to capture the participant's face, ensuring anonymity of participants and blinding of raters.  2.4.4 Data Processing  Kinematic Data Custom software written in Matlab (Version 14, The Mathworks Inc., Natick, MA, USA) was used to calculate kinematic rotation angles of the knee (cardan sequence: extension / adduction / internal rotation). Missing markers were recovered using a semiautomatic generalised cross variance B-spline fitting method within Matlab (5th order spline, 10Hz low pass filter).  110  The final kinematic data set was filtered using a  Butterworth filter (4th-order, zero-lag, low pass cut-off at 10Hz). Knee abduction was defined as tibial abduction relative to the femur. Knee extension, adduction and internal rotation angles were defined as positive values and were equal to zero in each participant when the long axes of femur and tibia were aligned, as defined during the initial stationary shot.  Lower limb reaching was derived using a novel calculation programmed in custom Matlab software. We defined reaching as the distance in the subject's frontal plane between the lateral malleolus and a point vertically below the greater trochanter. This variable was measured in three dimensions to account for rotation of the body out of the global frontal plane. The reaching distance was then normalised to the instantaneous leg length (distance from the greater trochanter to the lateral malleolus) to produce a ratio (FIGURE 14). We chose to examine the minimum reaching ratio value during 38  stance as this represented how close the participant came to achieving a vertical alignment of the lower limb during the side hop and side cut landings.  Greater trochanter •  .c s  C  Reach = Normalised reaching ratio Leg length  Lateral malleolus • Reach  FIGURE 14. Formula for normalised reaching ratio  Force Plate Data Force plate data enabled delineation of kinematic analysis to the stance phase. The stance phase, defined as the period from foot contact to toe off, was manually selected during Matlab processing from a graph of the vertical ground reaction force (VGRF) against time (FIGURE 15). The start of the stance phase was demarcated as the first data point rising above baseline and the end of stance as the last data point before the VGRF returned to baseline.  39  Date: 260E7 Subject: 3 Run: 40  Vertical ground reaction force  FIGURE 15. Vertical ground reaction force across stance phase  Digital Video Each participant's testing session was recorded on an individual digital video mini cassette and downloaded onto a standard personal computer using Microsoft ® Windows ® Movie Maker (Version 5.1, Microsoft Corporation). The video footage was then edited for viewing by raters and burnt onto compact disc (CD) using Microsoft ® Windows ® Media Player (Version 10, Microsoft Corporation).  2.4.5 Data Management Out of the nine trials performed by each participant, only three were included for kinematic analysis. Trials were excluded during Matlab processing if markers were missing at the start or end of the stance phase. Trials were also excluded if there was more than 10mm of motion between the markers on each limb segment as this marker 40  motion can significantly increase error. Finally, the kinematics of all remaining trials were time normalised to 100% of the stance phase and overlaid on a single graph. Through visual inspection, trials were excluded if they did not follow the same pattern of kinematic motion as the majority (FIGURE 16). After removing these outliers, each participant ended up with an average of seven drop jumps, seven side hops and 6.5 side cut trials. From the remaining trials, three were randomly selected for analysis. Only these trials appeared on the video shown to raters.  Adduction + ^ Int Rotation + Valgus + —  FIGURE 16. Selection of outliers from time normalised overlay plot (side cut)  41  2.5 Rating Protocol 2.5.1 Recruitment of Raters A pool of fifteen local physiotherapists with at least five years of experience in private practice and a high level of sports and orthopaedic expertise were contacted by email to request their involvement in the study as raters. Three female physiotherapists volunteered to participate. These individuals were contacted by email to further explain their roles and confirm their willingness to participate. After obtaining written and informed consent, the first of two rating sessions was scheduled.  2.5.2 Rater Training One week before the first rating session, raters were mailed a training video on CD and asked to view it at least once prior to attending. This twenty-minute rater training video was created by the investigator using Microsoft 0 Windows 0 Movie Maker. The video included background information about ACL injury risk and detailed instructions on how to rate the three different tasks. The video concluded with a practice rating session, during which raters were shown footage of individuals performing the tasks. This footage was left over from a larger study and did not include footage that would be viewed later in the rating sessions. The physiotherapists were asked to practice rating individuals based on what they had learned and were then provided with the 'correct' answers (operationalised using results from motion analysis).  42  2.5.3 Rating Procedures and Rules The first rating session, attended by all three physiotherapists and the investigator, ran for two hours and took place at the GF Strong Rehabilitation Research Laboratory. The video footage was projected onto a 2m x 2m screen in a darkened room. First, ten minutes were spent reviewing the rating instructions and practicing using footage left over from a larger study. Raters were permitted to share their decisions during practice time and if there were disagreements amongst raters on the methods used to classify individuals, the sources of these disagreements were discussed until a consensus could be reached.  The rating session commenced once all three physiotherapists felt confident with the rating instructions. All of the drop jumps were shown first, followed by the side hops and then the side cuts. This order was chosen to reflect increasing difficulty because it was expected that raters' observation skills would improve with practice. The forty participants were shown in a random order that differed for each task. For each task, individual participants performed three trials in succession. Raters were asked to give an overall rating of high risk or low risk based on the three trials viewed. In addition, the following rules were set: •  Raters were to complete their assessment in the 15 seconds provided between each participant.  •  They were permitted a single viewing only of each trial and the footage was never paused or rewound.  43  •  They were to refrain from sharing their ratings or making comments on the participants' performance to each other.  •  They were to focus only on the right lower limb for all three tasks.  •  For the drop jump, they were to focus only on the first landing from the box.  For the purposes of establishing intra-rater agreement, raters reassessed the same footage two weeks later. This time, raters were sent the footage on a CD and viewed it on their own personal computers. The raters also received an instruction pack to remind them of the rating guidelines and rules. The order of participants was unchanged from the previous rating session. Sessions were scheduled two weeks apart to reduce the likelihood of raters remembering their initial assessments. Anywhere from two days to two weeks has been suggested by Streiner and Norman as an appropriate retest interval for studies of this type.102  2.5.4 Development of the Risk Screening Guidelines Guidelines for risk screening were developed by the investigator, an orthopaedic physiotherapist, in consultation with an experienced sports physiotherapist. Thus far, no studies have attempted to qualitatively describe the exact position of the lower limb in which the ACL is at risk of injury. Therefore, it was necessary to create a novel set of observational risk screening guidelines. These guidelines were based on the current understanding of ACL injury risk factors and on the normal spectrum of lower limb biomechanics seen in their clinical practice. For example, while ideally an athlete would land from a hop with their hip joint centre vertically aligned over their knee and ankle 44  joint centres, in clinical practice this rarely occurs. Therefore, as can be seen below, if the participant was able to get their greater trochanter (which is lateral to the hip joint centre) over any part of the foot, they were given a low risk rating. It was also important that guidelines be based on the relative position of anatomical landmarks that would be easily discernible to raters during dynamic tasks.  The guidelines were designed to be as conservative as possible and capture as many potentially high risk athletes as possible. Therefore, if the raters detected any degree of knee abduction motion or lower limb reaching at all, they were asked to assign a high risk rating. Furthermore, if only one of the three trials was deemed high risk, raters were requested to assign an overall high risk rating to that participant. The rating instructions for each task are shown below in FIGURES 17-19.  2.5.5 Data Management Raters recorded their assessments on standardised rating sheets, an example of which is provided in APPENDIX VII. Following this, ratings were inputted into a spreadsheet (Microsoft Office Excel ®, 2003, Microsoft Corporation). For the purposes of data analysis, the numeral '1' was assigned to high risk ratings and '0' to low risk ratings. 2x2 contingency tables, such as those shown in APPENDIX XII, were then created to enable hand calculations of agreement and validity (formulae appear in APPENDICES XII and XIV)  45  2.5.6 Observational Risk Screening Guidelines  DROP JUMP RATING GUIDELINES "If the patella moves inwards and ends ^"If the patella lands in line with the big up medial to the big toe, rate the^toe rate the individual as individual as^  LOW RISK  HIGH RISK"  FIGURE 17. High risk and low risk drop jump landings  46  SIDE HOP RATING GUIDELINES "If the greater trochanter never makes it "If the greater trochanter gets over any over the foot, rate the individual as^part of the foot, rate the individual as HIGH RISK"  ^  LOW RISK"  FIGURE 18. High risk and low risk side hop landings  47  SIDE CUT RATING GUIDELINES "If the foot is placed laterally in relation ^"If any part of the foot is placed under to the greater trochanter, rate the^the greater trochanter, rate the individual as HIGH RISK"^individual as LOW RISK"  FIGURE 19. High risk and low risk side cut landings  2.6 Statistical Analysis 2.6.1 Test-retest Reliability 3D motion analysis is considered the 'gold standard' in the measurement of joint kinematics however it cannot be considered as such if it lacks adequate reliability. An unreliable or inconsistent gold standard would have been detrimental to the evaluation of validity in this study. To examine the test-retest reliability of motion analysis 48  procedures, ten participants performed the laboratory-based testing procedures twice, one week apart.  To find the most reliable way of measuring knee abduction, the test-retest reliability of two different variables were compared. These were 1) the peak knee abduction angle during stance and 2) the degree of abduction motion during stance. The peak angle was taken as the maximum 'absolute' degree of knee abduction reached during landing ('absolute' indicating that knee abduction was converted from a negative value to a positive value). The degree of abduction motion was calculated by subtracting the minimum absolute knee abduction angle during landing from the maximum absolute knee abduction angle (FIGURE 20). The measurement of lower limb reaching was also evaluated for reliability. This was of particular importance seeing as, to our knowledge, this was the first study to use this variable or to estimate its reliability.  In summary, the following variables were measured at both time points: 1) peak knee abduction angle on drop jump landing; 2) degree of knee abduction motion on drop jump landing; 3) lower limb reaching on side hop landing; and 4) lower limb reaching on side cut plant. Following removal of outliers using procedures outlined in section 2.4.5, a mean value was calculated for each participant from the remaining three trials.  The reliability of mean values was calculated using two methods: (1) intraclass correlation coefficients (model 3,k) (ICC3,k) and (2) Bland and Altman methods of agreement. 4 We also calculated the standard error of measurement (SEM) for each variable (see APPENDIX VIII for formula). 49  Knee Kinematics  Minimum knee abduction angle  -o  C  -20-  as  Maximum knee abduction angle  —4  0.1^0.2^0.3^0.4^0.5^0.6^0.7^0.8^0.9^1 Fraction of Stance  FIGURE 20. Maximum and minimum knee abduction across time normalised stance phase  1) The ICC3 k , was calculated using a repeated measures analysis of variance (ANOVA) design. The first integer in this model (3) indicates that the investigators in this study were the only investigators of interest and that the results would not be generalised to the larger population of investigators.83 The second integer (k) indicates that reliability analysis was performed using mean (rather than individual trial) values from motion analysis.83 ICC formulae appear in APPENDIX IX. Analyses were carried out in SPSS version 11.5 (SPSS, Inc., Chicago, USA).  50  Hypothesis: It was hypothesised that for all four variables (peak knee abduction, knee abduction motion, lower limb reaching on side hop and lower limb reaching on side cut),  ICCs would represent 'good' reliability ( ^0.75) and be significant at the .05 leve1.83  2) Bland and Altman methods estimate the agreement between values and are particularly useful for clinical measures.86 The differences between each participant's time one and time two means were plotted against the average of the two means. The mean of the differences and the 95% limits of agreement (2 x standard deviation of differences) were also calculated. These formulae appear in APPENDIX X. Bland and Altman plots were created using SPSS version 11.5 (SPSS, Inc., Chicago, USA).  Hypotheses: It was hypothesised that 1) the mean difference would be close to zero, 2) the data points would be evenly distributed about the mean difference line and 3) the limits of agreement would be within acceptable clinical ranges for each of the variables.  2.6.2 Rater Agreement  Intrarater agreement, was assessed using the kappa statistic (K). Kappa is a measure of the proportion of agreement within and between raters after chance agreement has been removed. It is considered the most appropriate method of analysis for agreement of categorical data.106 Kappa values were calculated for agreement within individual raters and also for a consensus rating. The consensus rating was determined by taking  the majority decision from the three raters. For example if two out of three raters labelled a participant as high risk, the participant received an overall rating of high risk. 51  To determine interrater agreement, the multirater kappa coefficient was calculated at time one and time two. This statistic, proposed by Fleiss, is used to estimate agreement between more than two raters.23 Agreement between rater pairs was also calculated at both time points using the standard kappa statistic. Agreement was examined between raters one and two, raters two and three and raters three and one.  For between and within-rater comparisons, the percentage of agreement was also calculated. These calculations and kappa analyses were performed by hand and verified using Analyse-it Software (Method Evaluation Edition, Analyse-it Software, Ltd., Leeds, England). Multirater kappa analyses were performed using web-based software based on equations presented in Fleiss (Copyright © 2004 Jason King, Ph.D. Available at http://www.ccit.bcm.tmc.eduaing/homepage/kappa.xls).23 A scale proposed by Landis and Koch was used to interpret the magnitude of agreement from a range of kappa values (APPENDIX XI).57 Kappa calculations and data contingency tables appear in APPENDIX XII. See APPENDIX XIII for a summary of assumptions that must be met in order to obtain valid kappa values.  Hypotheses: -It was hypothesised that for intrarater agreement, kappa values would be 'substantial' (^ 0.61) and significantly greater than zero at the .05 level (1-tailed). -It was hypothesised that for interrater agreement, multirater and standard kappa values would be 'substantial' ( ^ 0.61) and significantly greater than zero at the .05 level (1tailed).  52  2.6.3 Validity Validity was reported as sensitivity, specificity and positive and negative predictive values (PV+/PV-). Sensitivity represents the proportion of 'true positives' (ie the proportion of individuals with a high risk rating out of all of the individuals who are 'truly high risk'); specificity represents the proportion of 'true negatives' (ie the proportion of individuals with a low risk rating out of all of the individuals who are 'truly low risk'); PV+ represents the likelihood that a participant who tests positive actually has the disorder (Le. the likelihood that a participant with a high risk rating actually has high risk neuromuscular characteristics); and PV- represents the likelihood that a participant who tests negative does not have the disorder (ie. the likelihood that a participant with a low risk rating actually has low risk neuromuscular characteristics).  Gold Standard One: Motion Analysis In the absence of any high quality prospective studies that have determined the point at which the degree of knee abduction or lower limb reaching become injurious to the ACL, it was necessary to assign our own cut point between 'truly low risk' and 'truly high risk' groups with the help of an expert rater. The investigator, an experienced physiotherapist who co-developed the risk screening guidelines, acted as this expert.  The video footage of all 40 participants was reviewed by the expert rater. The rater was able to pause, slow down and rewind the video footage to allow precise application of the risk screening guidelines. Each participant subsequently received an 'expert rating' of high risk or low risk. This process was undertaken twice, one month apart. If there  53  were differences between time one and two ratings, the rater reviewed the footage a third time to come to a final decision.  Following this, a receiver operating characteristic (ROC) curve was constructed, comparing 'expert ratings' with values from 3D motion analysis. Sensitivity was plotted against 1-specificity for the entire range of kinematic values. For example, for knee abduction, the sensitivity and specificity of expert ratings were calculated as if the cut point was 4°. It was also calculated as if the cut point was 5°, 6° and so on, up to 29°. The same process was undertaken for each task so that for the side hop for example, sensitivity and specificity were calculated for the full range of reaching values from 0.16 to 0.43.  From the plots of sensitivity versus 1-specificity, a final cut off point was chosen for each task. It is recommended that the point on the curve closest to the upper left hand corner is chosen, provided that this point has an optimal balance of sensitivity and specificity.  101  This process was performed three times in total in order to obtain a  different cut point for each of the three tasks.  Individuals with kinematic values above the cut point were given a 'true' rating of high risk and those below the cut point were given a 'true' rating of low risk. Sensitivity, specificity, PV+ and PV- were calculated by comparing these 'true ratings' with physiotherapists' ratings.  54  Gold Standard Two: Expert Rater Sensitivity, specificity, PV+ and PV- were also obtained using an alternate gold standard measure. Instead of using 3D motion analysis as a gold standard measure of 'true risk', the expert rater was regarded as the gold standard. 'Expert ratings' were compared with physiotherapists' ratings to calculate alternate sensitivity, specificity, PV+ and PVvalues. These calculations were performed by hand and verified using Analyse-it Method Evaluation Edition (Analyse-it Software, Ltd., Leeds, England). The formulae for hand calculations are shown in APPENDIX XIV.  Hypotheses: It was hypothesised that observational risk screening would detect individuals with high risk neuromuscular characteristics with a high degree of sensitivity ( ^0.8) and moderate specificity (^0.5). These values were chosen based on ideal risk screening outcomes (see discussion, p81).  2.6.4 Relationship between Postural Measures and Knee Abduction  Postural measures of rear foot eversion, inversion, knee extension, knee flexion, navicular drop, knee valgus and knee varus were each compared with the degree of knee abduction using the Pearson product-moment coefficient of correlation. This analysis enabled detection of a potential relationship between static postural characteristics and dynamic knee abduction angles. Analyses were carried out in SPSS version 11.5 (SPSS, Inc., Chicago, USA).  55  Hypothesis: It was hypothesised that there would be no significant correlations at the .05 level (2-tailed).  2.6.5 Relationship between Gold Standard Variables In order to evaluate whether we needed to use all three tasks for risk screening, we looked for any redundancy between measures. First, we examined the correlation between lower limb reaching values for the side hop and for the side cut to determine whether individuals performed side hops and side cuts with a similar degree of lower limb reaching. We also examined the correlation between lower limb reaching values and knee abduction motion to determine whether individuals who reached more also landed from the drop jump with a greater degree of knee abduction. Statistical analyses were performed using SPSS version 11.5 (SPSS, Inc., Chicago, USA). A scale provided by Portney and Watkins was used to describe the magnitude of correlations (APPENDIX XV).83  Hypotheses: It was hypothesised that the correlations between side hop and side cut reaching values would be of moderate strength ( ^0.50) and significant at the .05 level (2-tailed) and that there would be no significant correlations at the .05 level (2-tailed) between reaching and abduction values.  Agreement was also examined between the expert ratings for different tasks. This analysis was performed using the kappa statistic and was aimed at establishing whether individuals received similar ratings from the expert regardless of the task being 56  performed. Kappa analyses were performed by hand and verified using Analyse-it Software (Method Evaluation Edition, Analyse-it Software, Ltd., Leeds, England).  Hypotheses: It was hypothesised that agreement between expert ratings on the side hop and side cut would be 'substantial'  (K  ^0.61) and significantly greater than zero at the .05 level (1-  tailed). It was hypothesised that agreement between expert ratings on the drop jump and side hop, and on the drop jump and side cut, would be 'slight'  (K  ^0.21) and not  significantly greater than zero at the .05 level (1-tailed).  2.6.6 Sample Size Calculations Sample size calculations were based on the minimum number needed to detect a statistically significant kappa coefficient (p<.05). Using Donner and Eliaszew's goodness-of-fit formula,16 Sim and Wright provided estimated sample sizes for a range of different requirements.98 Their calculations were based on a study involving two raters (or two rating sessions) and a dichotomous rating scale. At 80% power and 95% confidence, a minimum of 18 participants are required to detect a kappa of 0.60 or more and a minimum of 39 participants are required to detect a kappa of 0.40 or more. This assumes a one-tailed test and a null hypothesis of zero. To optimise power, we selected a sample based on the lower kappa value of 0.40. Therefore, a total of 40 participants were included in this study. This was considered a conservative estimate as these calculations were based on a study with two raters. Having more than two raters further  57  reduced the number of participants needed to obtain optimal power. Sim and Wright's table of sample sizes is reproduced in part in TABLE 1 below.  TABLE 1. Sample size requirements for different values of kappa Kappa to detect  Sample size required*  0.40  39  0.50  25  0.60  18  0.70  13  0.80  10  0.90  8  *NB this represents the n needed assuming a 1 tailed test, a null value of 0, alpha <.05 and power at 80%.  58  3 RESULTS 3.1 Participant characteristics Forty female soccer players between the ages of 13 and 17 years participated in this study. 30 participants were recruited from national level teams and ten from gold level teams. Their characteristics are shown below in TABLE 2. All forty participants selected the right leg as their dominant leg. Therefore biomechanical and postural analyses were performed on the right lower limb only.  TABLE 2. Participant characteristics SD  Range  1  13-17  N  Mean  Age (years)  40  15  Height (m)  40  Weight (kg)  40  BMI (kg/m2)  40  Right Rear Foot Eversion (*)  33  5  Right Rear Foot Inversion (*)  1  2*  Right Navicular Drop (cm)  39  0.5  0.3  0.2 1.0  Knee Valgus (cm)  21  1.9  1.6  0.5 6.0  Knee Varus (cm)  18  2.1  1.4  1 5  Right Knee Hyper-extension (*)  16  4  2  2 8  Right Knee Flexion (*)  11  3  2  2 6  0.06  1.55-1.85  60.0  8.5  43.8-78.0  21.9  2.3  17.3-26.1  1.65  2  2 10 -  -  -  -  -  -  *Because n=1 for this variable, this is the actual value for this subject  59  3.2 Test-retest reliability 3.2.1 ICCs  Ten participants completed the laboratory based testing procedures on a second occasion, approximately one week after the first testing session. Based on ICC3,k values, test-retest reliability was better for knee abduction motion (0.76) than peak knee abduction (0.59) and was better on reaching for the side cut (0.94) than the side hop  (0.76). ICCs for the four motion analysis variables of interest are shown in TABLE 3.  TABLE 3. Test-Retest Reliability of Motion Analysis Variables — ICC3,k Values  ICC3,k Value  95% Confidence Intervals  Maximum Knee Abduction Angle  0.59  -0.67 to 0.90  Knee Abduction Motion  0.76*  0.02 to 0.94  Normalised Minimum Reaching Ratio (Side Hop)  0.76*  0.05 to 0.94  Normalised Minimum Reaching Ratio (Side Cut)  0.94*  0.74 to 0.98  * Indicates kappa significantly greater than zero (p<.05); ICC values ^0.75 are in bold  3.2.2 Bland and Altman Methods  Bland and Altman plots appear below for each of the 4 variables. For peak knee abduction (FIGURE 21), the mean difference between time one and time two was close  to zero (1.64'). All but one data point fell between the limits of agreement and the data points were fairly equally distributed above and below the mean difference line (six below, four above). The limits of agreement were up to ±12.5° for this variable. For knee abduction motion, the mean difference was also close to zero, this time with a 60  slightly negative bias (-1.53°) (FIGURE 22). All of the data points fell between the limits of agreement and again, the data points were evenly distributed about the mean difference line (five below, four above and one on the line). For this variable the limits of agreement were tighter, with the differences between time one and time two being no greater than ±8°. With a higher ICC value and tighter limits of agreement, knee abduction motion (maximum-minimum angle) proved to be a more reliable variable than the peak knee abduction angle (maximum angle).  For the lower limb reaching ratio, the mean difference was just slightly above zero for the side hop (0.02) (FIGURE 23). All of the data points fell between the limits of agreement and the data points were fairly evenly distributed above and below the mean difference line (four below, six above). The limits of agreement were up to ±0.08 for this variable. For the side cut, the mean of the differences between time one and time two was slightly below zero (-0.02) (FIGURE 24). All of the data points fell between the limits of agreement and the data points were again evenly distributed about the mean difference line (five below, four above and one on the line). This time the limits of agreement were tighter, with the differences between time one and time two being no greater than ±0.04. The ICC value was also higher for the side cut, indicating that for this task, lower limb reaching could be measured with better reliability.  61  Mean difference 95% limits of agreement  Drop jump Maximum knee abduction angle - means 20 3  15  a  4 a  7.  5 5  10  8 02 ^ a  ^  a  a^  9  6 a -10 -15 -20 -24  -22^-20^-18^-1-6^-14^-12^-10  BRMean (degrees)  FIGURE 21. Bland and Altman Graph for Maximum Knee Abduction Angle  Mean difference 95% limits of agreement  Drop jump Max-min knee abduction angle - means 20 15 10  a)  2  a 5  9 a  a)  -cs^0  o  4 7^  1 -5  5 a  3 a  2 a^8 10  -10 -15 -20 4  6  8  10  12  14  16  11  BRMean (degrees)  FIGURE 22. Bland and Altman Graph for Knee Abduction Motion  62  Side hop  Mean difference 95% limits of agreement  Minimum normalised reaching value - means .20 .15 .10 .05  6  7 10 3  0  0.00  2 50  CO  4  -.05 -.10 -.15 -.20 22  .24  .26  .28  .30  .32  .34  .36  .38  BRMean  FIGURE 23. Bland and Altman Graph for Lower Limb Reaching (Side Hop)  Unanticipated side cut Minimum normalised reaching value - means  Mean difference 95% limits of agreement  .20 .15 .10 .05  8  tO  TS ^  0.00  20  CO -.05  R a  1  3 4  9  ^  -.10 -.15 -.20 22^.24^.26^.28^.30^.32  ^  34  ^  36^.38  BRMean  FIGURE 24. Bland and Altman Graph for Lower Limb Reaching (Side Cut)  Abbreviations: BRDiff, Difference between baseline and retest values; BRMean, Mean of baseline and retest values 63  3.2.3 SEM  The standard error of measurement was 3.930 for the peak knee abduction angle, 2.600 for knee abduction motion, 0.02 for side hop reaching and 0.01 for side cut reaching. Because the data for these variables were not normally distributed in this sample of ten, SEM values must be interpreted with care.  3.3 Rater agreement The three raters were female physiotherapists with between 8.5 and 14 years (mean ± SD, 12.17 ± 3.18) of clinical experience in sports physiotherapy and orthopaedics. At time one and time two, they rated each of the participants as high risk or low risk for each of the tasks. Ratings were fairly evenly split between high risk and low risk for the drop jump. There were more high risk than low risk ratings for the side hop and the side cut. The high risk/low risk rating frequencies appear in APPENDIX XVI.  3.3.1 Intrarater Agreement (TABLE 4)  It was hypothesised that intrarater agreement would be 'substantial' with kappa values ^0.61. For the drop lump, the hypothesised level of agreement was met; with individual raters attaining 'substantial' to 'almost perfect' kappa coefficients of 0.75 to 0.85. There  was 'almost perfect' agreement between consensus ratings at time one and time two, with a kappa coefficient of 0.85.  64  TABLE 4. Intrarater Agreement  Drop Jump  Side Hop  Side Cut  * Indicates  K  Percentage of Agreement  Kappa Value  Rater 1  90.0%  Rater 2  95% Confidence Intervals Lower  Upper  0.80*  0.65  1.00  92.5%  0.85*  0.72  1.00  Rater 3  87.5%  0.75*  0.58  1.00  Consensus  92.5%  0.85*  0.71  1.00  Rater 1  92.5%  0.79*  0.61  1.00  Rater 2  92.5%  0.83*  0.67  1.00  Rater 3  97.5%  0.94*  0.85  1.00  Consensus  92.5%  0.83*  0.67  1.00  Rater 1  82.5%  0.31*  0.02  1.00  Rater 2  82.5%  0.58*  0.34  1.00  Rater 3  90.0%  0.75*  0.56  1.00  Consensus  92.5%  0.79*  0.61  1.00  significantly greater than zero (p<.05);  K  values ^0.61 are in bold  For the side hop, intrarater agreement was slightly higher, with kappa coefficients for individual raters ranging from 0.79 to 0.94. There was 'almost perfect' agreement between consensus ratings at time one and time two, with a kappa coefficient of 0.83. For the side hop, hypothesised levels of intrarater agreement were met.  For the side cut, intra-rater agreement was lower, with kappa values for individual raters ranging from 0.31 to 0.75. Only one of the raters met the hypothesised level of agreement with this task. For the consensus rating, a kappa coefficient of 0.79 was attained, meeting the hypothesised level of agreement.  65  3.3.2 Interrater Agreement (TABLE 5) lnterrater agreement was examined at both time points. For the drop jump, the  multirater kappa coefficient exceeded the hypothesised value of 0.61 at both time points (0.80 and 0.77). Kappa coefficients for rater pairs also reached the hypothesised value, ranging from 0.70 to 0.90.  For the side hop, the multirater kappa value was lower than for the drop jump but still managed to meet hypothesised targets (0.72 and 0.64). Kappa coefficients for rater  pairs ranged from 0.46 to 0.77, with only one of the pairs failing to reach hypothesised targets.  For the side cut, the multirater kappa coefficient was lower still and fell from 'moderate' (0.50) at time one to 'fair' (0.24) at time two. These values did not reach the hypothesised value. None of the rater pairs attained the hypothesised level of agreement at either time point, with kappa values ranging from 0.06 to 0.60.  3.4 Cut Off Points To obtain an optimal cut off point for discrimination of 'truly low risk' and 'truly high risk' individuals the video trials were shown to an expert rater. Ratings were fairly evenly split between high risk and low risk for all three tasks. The high risk/low risk frequencies of 'expert ratings' appear in APPENDIX XVII.  66  Ta  to  LO 0 0 0 CO^0 0 0) 0 0 0 CO^0 0 0^-,---^<-0 0 0 0.9 0 0  6 6 6 6 6^6 c?  0) CO 0 CV 0)^CO CV LO '1 CO CV CO^CV <-  d  tr)^h.^03 I,-^CO^ CO`,1CN 7:t 0 CV h- CO ih•^(1;)^  fa;^O a) 6 ci 6 6 6 6  010^LC)^U?  N: 0 N.-^N.: LU C N.1 00 0) 03^1"--- CO CO  1(  81:•' LC) 0 1-.0^Lc) 0 N- 0 h:^CV h.: 0 NCO 0) h-^th•  *^*^1,^*^*^*^*^* 10 000 Is* 0 (NO) 0 0.1 is-^I,.^0)^CO^r••••^f•-•^CO^(S)^1.0 00000000 C) CC) C)  0 LC) LC) LO CV CO 0) CO  Q  67  The sensitivity and specificity of expert ratings were calculated using a range of knee abduction motion values as cut off points. Receiver operating characteristic (ROC) curves, plotting sensitivity against 1-specificity are shown in APPENDICES XVIII, XIX and )0( for each of the tasks. A cut off point of 10.83° of knee abduction motion was chosen for the drop jump. Out of all points closest to the upper left hand corner of the plot, this cut off point had the highest sensitivity (0.87). A reaching ratio of 0.29 was chosen as the cut off point for the side hop. While there was one point closer to the upper left hand corner than this one, the chosen cut off point had higher sensitivity (1.00) and specificity (0.89). A cut off point of 0.32 was chosen for the side cut. This cut off point was again chosen based on its high sensitivity value of 0.94 and specificity of 0.71. Participants were assigned to 'truly high risk' and 'truly low risk' groups based on these cut off points. The frequency and prevalence of these 'true ratings' appear in APPENDIX XVII.  3.5 Validity of Observational Ratings- Method 1  (TABLE 6)  Sensitivity, specificity, PV+ and PV- were calculated at both time points by comparing physiotherapists' ratings with 'true ratings'. It was hypothesised that sensitivity would be ^0.80 and that specificity would be ^0.50.  3.5.1 Sensitivity Sensitivity values ranged from 0.67 to 0.87 for the drop jump and hypothesised targets of ^0.80 were only met twice - by rater three at time one and rater one at time two.  68  I-  i=  i=  a,  csi  I.-  i=  Ta csi  a)  cc) C)  i= cs, i=  o co^cs) co O 000  03 co I-- r-- co 6 6 6 6  O  (0 0) LO CO IN- CO  0 00 0 0000  0) 0 0 0 CO 0 0 co  O  0 CV C (0 1.0 LO  o^o o 0• 0)0C o  0 CV CO 0 00 CO 0  LO CO  o 00  CO^Cs.1^C^C \ I,- CO CO co  c;  CA 00 cr)  6 00  000  (0 (0(0 CV Cr) c0 CO •zr  co;  •dr^'cl" a) a) 0 r-  0 et CO 0 a:: CI! 0 0 e-  O  •  O  O 0  000  10^Sr) 11,  000  6 6  6 6 6 cp  d  CO Ir) CO  0 •  CO N N N  (0 000 0  0000 00. 0. 0  O  co 000  CO 0 CNI CO 0 In '71' co  r-000  O  IN- CA I.'TN (0 l•-• CD  • (.0^,-LO LC) LO  •tt CO N N CO CO N. P-O 00  C'.^0 N (.l CO N-  N- Ce)  0  co co r-6 6 o  CO  0  0") CO N- CO CO 0  O  a)  .E  i= ‘a) i=  a) i=  i=  I-  (.0 CO (0(0  0 0 o  (N 00) (0 0 oo (0  •  0  000  •  •,zr  6 6  00) CO 0 CO  ci  10 141  e- CO 0 00 CO  e-^CO CO 0 CA  69  The desired level of sensitivity was reached for the consensus rating at time one, with a value of 0.80. At time two, sensitivity was lower (0.73) and did not meet the hypothesised value.  For the side hop, all of the raters achieved perfect sensitivity except for rater one at time two, who achieved a sensitivity value of 0.96. The consensus rating also achieved a sensitivity of 1.00 at both time points.  For the side cut, individual raters did not perform quite as well as they did on the side hop. Rater one achieved a sensitivity of 1.00 at both time points but raters two and three had sensitivity values between 0.88 and 0.94. Nonetheless, their values still exceeded hypothesised levels. The consensus rating also achieved a sensitivity of 1.00 at both time points.  3.5.2 Specificity For the drop jump, specificity for individual raters met hypothesised values of ?.0.50 at both time points, with values ranging from 0.60 to 0.72. For the consensus rating specificity was equal to 0.72 at both time one and time two.  For the side hop, rater one was the only individual rater to not meet hypothesised specificity targets, with a value of 0.44 at time two. The remaining raters attained  70  specificity values between 0.56 and 0.79. The consensus rating met hypothesised specificity levels at both time points with values of 0.72 and 0.67.  The side cut had the lowest specificity values of any of the tasks and for the most part, did not meet hypothesised levels at either time point. Only one of the individual raters met the hypothesised specificity level at time two, with a value of 0.50. The remaining raters had specificity values ranging from 0.08 to 0.42. Specificity values for consensus ratings were 0.38 and 0.42.  3.5.3 Positive and Negative Predictive Values (PV+/PV-) For the drop jump, raters achieved positive predictive values ranging from 0.52 to 0.65. Negative predictive values were lowest for this task, ranging from 0.77 to 0.90.  At both time points, PV+ was highest for the side hop, with individual and consensus values ranging from 0.68 to 0.85. The side hop also had the highest PV- at both time points with values ranging from 0.89 to 1.00.  Positive predictive values were lowest for the side cut, ranging from 0.42 to 0.56. Negative predictive values were high, with individual values ranging from 0.83 to 1.00 and consensus values being equal to 1.00 at both time points.  71  3.6 Validity of Observational Ratings- Method 2 (TABLE 7) Validity was also determined using an alternate method of comparing physiotherapists' ratings to 'expert ratings'. Using this method, sensitivity and specificity improved for the  drop lump, with values now exceeding hypothesised levels at both time points. Sensitivity improved from 0.80 to 0.91 at time one and 0.73 to 0.81 at time two. Specificity improved from 0.72 to 1.00 at time one and 0.64 to 0.95 at time two. There were also slight improvements in the predictive values.  For the side cut and the side hop, sensitivity and the PV- did not change for the most part but there were slight improvements to specificity and the PV+.  3.7 Relationship between Postural Measures and Knee Abduction For all forty participants, Pearson correlations were calculated between mean knee abduction values and selected postural measures. Correlations ranged from 0.03 (navicular drop) to 0.21 (knee valgus). In support of the stated hypothesis, none of these correlations were significant at the .05 level (2-tailed) (APPENDIX XXI). Out of all postural variables, only rearfoot eversion and navicular drop data were normally distributed. Thus, caution must be applied in the interpretation of the results for the other non-normally distributed variables. In light of this non-normal data, the same correlations were examined by calculating Spearman's rho. Similar results were obtained to those above and appear in APPENDIX XXI.  72  3.8 Relationship Between Gold Standard Variables While knee abduction motion and side cut reaching values were normally distributed in this sample, reaching values for the side hop were not (APPENDICES XXII-XXIV). The following results relating to side hop data must be therefore be interpreted with caution. There was a weak, positive correlation (0.34) between lower limb reaching values for the side hop and the side cut. This correlation was significant at the .05 level (2-tailed). There were weak, negative correlations between lower limb reaching and knee abduction values. The correlation with side hop reaching values was -0.30 and with side cut reaching values was -0.33. These correlations were significant at the .05 level (2tailed).  There was a 'fair' level of agreement between expert ratings for the side hop and side cut, with a kappa value of 0.39. This kappa value was significantly greater than zero at the 0.05 level (2-tailed) but did not meet hypothesised levels. Kappa values for agreement between the drop jump and side hop and between the drop jump and side cut were negative, representing worse than chance agreement (-0.16 and -0.06, respectively). This was in support of the stated hypothesis.  73  4 DISCUSSION The purpose of this study was to examine physiotherapist agreement and validity in evaluating ACL injury risk using observational risk screening guidelines. Three physiotherapists viewed video footage of 40 female soccer players who performed three tasks - a drop jump landing, a side hop and an unanticipated side cut. Drop jump landings were rated based on the degree of dynamic knee abduction. Side cuts and side hops were rated based on the degree of lower limb reaching.  4.1 Test-Retest Reliability Ten individuals performed the laboratory-based testing procedures twice, one week apart. Test-retest reliability analyses were performed on four key motion analysis variables: peak knee abduction angle, knee abduction motion, lower limb reaching (side hop) and lower limb reaching (side cut). Bland and Altman methods were used to detect bias in our measurements and estimate the extent of disagreement (limits of agreement') between scores. ICCs were used to assess both correlation and agreement between scores.83  The ICC for peak knee abduction did not meet the hypothesised value of ^0.75. Therefore, this variable was discarded in favour of knee abduction motion which had better reliability and an ICC of 0.76. In further support of this decision, the limits of agreement were almost 50 tighter for knee abduction motion than peak knee abduction.  74  Based on their ICCs, reaching values for both the side hop and the side cut were measured with adequate reliability. However, side cut reaching values were more reliable than side hop reaching values, with ICCs of 0.94 compared to 0.76. The limits of agreement were also slightly tighter for the side cut than the side hop.  Bias is said to be present when one set of measurements are systematically higher or lower than another set. Based on a mean difference close to zero and the even distribution of data points around the mean difference line, we were able to conclude that there was minimal bias for all four variables.  4.1.1 Possible Factors Affecting Reliability Much of the error in 3D motion analysis is thought to arise from the movement of surface markers which is accentuated with more dynamic tasks. Seven different surface markers were used to derive knee abduction values and each of these markers may have contributed to movement-related error. Because the side hop was a more dynamic task than the side cut, its poorer reliability may have also resulted from marker motion. Nonetheless, the reliability for both the side hop and side cut tasks was within acceptable limits. Unlike knee abduction values, lower limb reaching values were derived from only two surface markers which may have helped lessen the influence of marker-related error.  The use of intra-cortical bone pins instead of surface markers can significantly reduce error. In a study of running kinematics by Reinschmidt et al, there were differences in 75  knee abduction angles of up to 11.10, depending on whether intra-cortical or surface markers were used.88 In an attempt to lessen the influence of marker motion in our study we excluded trials with excessive marker motion (>10mm). While this may have lessened the degree of error somewhat, this source of error was not completely eliminated. We were also unable to completely control other sources of error such as inconsistent task performance or inconsistent marker placement.  4.2 Agreement Intrarater and interrater agreement for the drop jump and side hop were relatively high in this study compared to others. In a study by Chmielewski et al, three clinicians used observational rating guidelines to assess the degree of frontal plane trunk and lower limb motion exhibited during a unilateral squat and a lateral step-down task.9 Intrarater weighted kappa coefficients ranged from 0.13 to 0.68 and interrater multirater kappa coefficients ranged from 0.01 to 0.22. Raters used two different rating methods to classify participants - an overall and a specific method. The authors felt that their criteria for both rating methods were too ambiguous, with the overall method consisting of the categories 'poor', 'fair' and 'good'; and the specific method consisting of the categories 'excessive', 'moderate', 'small' and 'no deviation from neutral'. For future studies they recommended developing more explicit rating criteria incorporating anatomical landmarks, such as "knee moves to the inside of the great toe".9  (p129)  We followed their recommendations and based our screening guidelines on anatomical landmarks that would be easily discernible to raters, such as the patella, the big toe and  76  the greater trochanter. Consequently, for the drop jump and the side hop, all three raters met hypothesised levels of intrarater agreement (K ^0.61) with kappa values ranging from 0.75 to 0.94. lnterrater (multirater) kappa coefficients also exceeded hypothesised values (K^0.61) for the drop jump and side hop, ranging from 0.64 to 0.80.  In a study by Hickey et al, 11 physiotherapists rated the shoulder and scapular movements of nine subjects.43 Only five out of 20 subjects had two or more physiotherapists agree on the type of anomalous movement occurring in their shoulders and kappa values ranged from 0.17 to 0.34. In the studies by both Hickey et at and Chmielewski et al, multiple scoring response categories were provided to raters. Chmielewski et al acknowledged that rater agreement would have been better had the number of rating categories in their study been limited to two as the presence of a middle category reduced the opportunity for agreement.9  To facilitate scoring and enhance agreement, our study incorporated a dichotomous scale (high risk versus low risk). Restricting the number of rating categories to two helped us to achieve high levels of agreement. However, it must be acknowledged that a dichotomous scale does limit the level of detail provided. Certainly, when participants had subtle degrees of knee abduction, raters had more difficulty, which was reflected in poorer agreement for these participants. Raters may have been able to make use of a third 'moderate risk' category in these situations. Nonetheless, these rating guidelines were designed to be used as a preliminary screening tool. Providing that a sufficient level of sensitivity is reached, a front line preliminary screening tool does not need to be overly detailed. More specific and detailed athlete assessment can be performed 77  following identification of athletes clearly at risk. Furthermore, the uncomplicated nature of the rating guidelines makes this method highly user-friendly and optimises the likelihood that clinicians and coaches will start implementing risk screening on a more routine basis.  4.2.1 Possible Factors Affecting Agreement Our results suggest that physiotherapy experience may not have any effect on rating skill, with the two most experienced physiotherapists having both the highest and the lowest degree of intra-rater agreement. While it must be acknowledged that this observation is based on a small sample of only three physiotherapists, it is supported by previous studies. Knudson compared the observational rating expertise of kinesiology professors with extensive teaching and coaching experience and kinesiology students.54 The raters were asked to estimate the degree of trunk lean and knee flexion of subjects landing from a jump. The investigator found that the experience of the professors did not improve their rating ability, with the professors actually having poorer reliability and validity than the students. Similar findings were reported by Somers et al who investigated the influence of experience on the reliability of observational measurements of forefoot position.101 Two clinicians with more than ten years of experience and two physiotherapy students were asked to estimate the forefoot position of ten participants. They found no significant difference in the intrarater or interrater reliability between experienced and inexperienced clinicians. It has been suggested that other factors may be more influential than experience, including dynamic visual acuity, experience in performing the task being rated, the amount of specific training received and the perceptual style of the raters.54 78  Perceptual rating style can lead to rater bias, where one rater has a systematic tendency to favour one rating category. Bias will have a negative influence on kappa values. At time two, rater one had a bias towards high risk ratings, labelling only two participants out of forty as low risk on the side cut task. The bias greatly reduced her intrarater kappa value. Despite achieving the same percentage of agreement as rater two, rater one's kappa value for the side cut was approximately half of rater two's. This bias also affected interrater agreement. At time two, despite attaining more than 70% agreement, rater pairs that included rater one had kappa values as low as 0.06. While rater one had been practicing fewer years than the other two raters, it is more likely that she merely interpreted the screening guidelines differently to the other raters.  This highlights the importance of consistent and thorough rater training. In a study by Eastlack et al, 54 physiotherapists performed observational gait analysis assessments on three subjects. 18 Multirater kappa coefficients for interrater agreement ranged from 0.11 to 0.52. The authors alleged that the poor agreement between raters resulted from inconsistent interpretation of the assessment method and that greater standardisation of gait analysis training was needed.  In our study, the use of different training methods at time one and two may have affected rater agreement. At time one, raters received ten minutes of rater training immediately prior to rating the participants. They did not receive this training prior to the second session as it was assumed that they would remember the guidelines from time one. As a result, interrater agreement was lower at time two than time one for all three  79  tasks. While raters were reminded of the screening guidelines, it is possible that at time two they should have received a more detailed review of the instructions.  lntrarater agreement was calculated for the consensus and well as for individual ratings. The consensus rating was determined by taking the majority decision from the three raters and it was expected that this would result in better agreement than that of individual raters. While this was true for certain outcomes, such as intrarater agreement for the side cut, this was not a consistent finding overall. Therefore, in practice, deciding whether to take a group consensus or to rely on the opinion of a single rater appears to be dependent upon how difficult the task is to rate and also upon the expertise of individual raters.  Hypothesised levels of intra and interrater agreement were met for the drop jump and the side hop. However, this was not the case for the side cut. Adequate interrater agreement was not obtained at either time point and only one of the raters obtained adequate intrarater agreement, suggesting that for this task, it was harder for the raters to be consistent. On examining the video footage of side cut trials, we observed several participants turning their bodies and lead legs towards the ball as they stepped forward. Invariably, there were disagreements within and/or between raters on these participants. Turning the leg inwards made it harder to see the lateral edge of the greater trochanter which was a key landmark in assessing the degree of reaching. In practice, it may be necessary to inform athletes that they must keep the toes of their lead leg facing forward during the side cut. However, it must be admitted that, during data collection, we found this hard to enforce. 80  Other authors have commented on the difficulties they have experienced with tasks similar to the side cut. In the study by McLean et al that compared 2D and 3D assessment methods, the 2D method was found to be unreliable with the shuttle run task where athletes had to initiate a large direction change after contact with the force plate.69 The authors noted that their frontal plane analysis methods were most amenable to movements in which the participant remained facing the camera throughout stance. In our study, this was a characteristic of the drop jump and side hop, but not the side cut.  4.3 Validity The perfect screening test has sensitivity of 1.0 and a specificity of 1.0. In reality however, most tests are imprecise to a certain degree and researchers are usually forced to sacrifice sensitivity for specificity or vice versa. It is up to the researcher to decide on the better outcome for participants and to set hypothesised sensitivity and specificity values accordingly.  Attaining a high sensitivity value minimises the number of 'false negatives' in a sample. False negatives are 'truly high risk' individuals who have been incorrectly labelled as low risk. In risk screening, it is important to minimise false negatives as these individuals may miss out on vital injury preventative initiatives and go on to experience an injury. To minimise the number of false negatives in this study, a high sensitivity value of ^0.80 was hypothesised. We decided it was more important to achieve high sensitivity values than high specificity values and therefore a lower specificity value of ^0.50 was  81  hypothesised. The main drawback of attaining a low specificity value is that some individuals will be incorrectly labelled as high risk. Providing there is no negative stigma attached to a label of high risk, the only disadvantage is that these 'false positives' might receive unnecessary injury prevention training.  Hypothesised sensitivity values were obtained for the side hop and the side cut but for the drop jump they were not. When sensitivity values were calculated for the consensus rating, rather than for individual ratings, the desired sensitivity levels were met for the drop jump at time one but were still insufficient at time two. For the side hop and the side cut, the consensus ratings achieved 100% sensitivity at both time points. This meant that all 'truly high risk' individuals were correctly identified by the raters and there were no false negatives.  It was surprising to observe that sensitivity values were generally lower at time two. However, in hindsight, this may have resulted from the different way in which the two rating sessions were run. The time one rating session was held at our research laboratory where the raters viewed the footage on a large screen in a quiet, darkened room. The projected image created an almost life-sized image of the participants and put the lower leg at the raters' eye-level. At time two raters were mailed the footage on CD and viewed it on their own home computers. Viewing conditions may have been suboptimal in this setting and raters may have had to contend with uncontrolled environmental distractions.  82  Hypothesised specificity levels were met by all three raters for the drop jump task alone. For the side hop, rater one's specificity at time two was insufficient owing to her having labelled over half of the 'truly low risk' individuals as high risk. Her bias towards high risk ratings was discussed previously. However, when the consensus rating was considered, specificity was acceptable for both the drop jump and the side hop. Out of all tasks, the lowest specificity values were obtained for the side cut. None of the raters met hypothesised targets at either time point. This was partly a result of the prevalence of low risk versus high risk individuals in our sample. For the side cut, there were 16 'truly high risk' individuals and 24 'truly low risk'. Because raters did not increase their low risk ratings accordingly, a large proportion of 'truly low risk' individuals were labelled as high risk. Thus, the number of false positives increased and specificity decreased.  At both time points, positive predictive values were highest for the side hop (0.68 to 0.85) and lowest for the side cut (0.42 to 0.56). Positive predictive values estimate the likelihood that an individual who has received a high risk rating is 'truly high risk'. The positive predictive value is an important guideline in deciding whether to implement a risk screening program. A PV+ of 0.50 indicates that half of the individuals with a high risk rating are not actually at risk. If further efforts are not put into discovering how many of these individuals are truly at risk, some will receive unnecessary injury prevention training which could be seen as a waste of resources.  A bias towards one rating category can have a negative effect on predictive values. For both the side hop and the side cut, raters categorised approximately three times more  83  individuals as high risk than low risk. Had raters been less cautious and rated fewer individuals as high risk, positive predictive values may have been higher for these tasks.  The side hop had the highest negative predictive value at both time points (0.89 to 1.00) and the drop jump had the lowest (0.77 to 0.90). PV- estimates the likelihood that an individual with a low risk rating is 'truly low risk'. For the side hop and the side cut, the consensus rating had negative predictive values equal to 1.00 at both time points. This indicates that 100% of the individuals who were rated as low risk were 'truly low risk'.  4.3.1 Possible Factors Affecting Validity The prevalence of 'truly high risk' and 'truly low risk' individuals was dependent on the cut off point chosen. Had the cut off point been lower for the drop jump and the side cut, there would have been more even numbers of high risk and low risk individuals. While this would have improved specificity and PV+ values, sensitivity would have been lower. For this study because we decided that sensitivity should be prioritised over specificity, cut off points were deliberately set high. Further prospective research needs to be done to determine what the actual biomechanical cut points are between individuals who are and those who are not at risk of ACL injury.  Because we still do not know how much reaching or knee abduction is too much, our cut off points were selected with the aid of ROC curves. The ROC curves plotted sensitivity against 1-specificity for a range of motion analysis values. For the side hop, the ROC curve followed an ideal pattern, starting from the lower left hand corner and 84  ascending straight up and across to the upper right hand corner. From this plot, the choice of a cut off point was clear, being the point closest to the upper left hand corner with the best sensitivity. The ROC curve for the side cut was less smooth, indicating poorer equivalency between the expert ratings and the 3D motion analysis values. The ROC curve for the drop jump was quite erratic and actually changed direction at lower cut point levels. As a result, it was more difficult to choose an ideal cut off point for this task.  The abnormal shape of the drop jump ROC curve resulted from the poor equivalency between expert ratings and the results from our 'gold standard' - 3D motion analysis. In accordance with the observational risk screening guidelines, individuals who landed with the patella medial to the big toe received a high risk rating by the expert. However, many of the individuals who received high risk expert ratings recorded unexpectedly low knee abduction values. It seems that what was observed by the expert rater in respect to frontal plane motion was not closely reflected in the results of 3D motion analysis.  Laboratory-based measurement error may have been partly responsible for this. However, it may also be that 3D measures cannot be used interchangeably with 2D measures (ie. ratings from video). McLean et al attempted to compare 2D and 3D measures of knee abduction with tasks such as a side-step cut and a side jump.69 They used video footage to measure the degree of knee abduction in the frontal plane and compared these values to those measured using 3D motion analysis. Within subjects, there was only a moderate correlation between 2D and 3D values (r2=0.25-0.36). Also, in agreement with our findings, their 3D angles were consistently smaller than the 85  corresponding angles measured from video. McLean et al hypothesised that this was due to combined knee flexion and hip internal rotation giving the appearance on video of a larger knee abduction angle. This is understandable, considering that the knee flexion and hip internal rotation that produces knee abduction will be more obvious when viewed in the frontal plane than in the orientation of the knee joint axis.  A poor association between 2D and 3D measures was also reported by Krosshaug et al.55 In this study, raters estimated lower limb kinematics from video footage of side cut trials. Raters were asked to classify a change in knee angle as `valgus', 'neutral' or `varus' between two predetermined time points on the video. These ratings were compared to the actual change detected with 3D motion analysis. For the side cut, video estimates and 3D knee abduction angles were found to have 'poor' agreement, with a kappa value of 0.19.  To address the poor compatibility of our drop jump ratings with 3D motion analysis, validity was also investigated using the expert, rather than 3D motion analysis, as the 'gold standard'. When this method was used, the sensitivity of drop jump ratings improved to the extent that hypothesised sensitivity targets were now reached at both time points. Using this method, hypothesised sensitivity values were met for all three tasks at both time points and hypothesised specificity values were met for all three tasks at time one and for the drop jump and side hop at time two.  86  4.4 Selection of Tasks for Risk Screening In this study, three different tasks were examined as potential components of an ACL risk screening program. When individual raters were considered, none of the tasks met hypothesised targets for all four outcomes. The drop jump failed on sensitivity, the side hop failed on specificity and the side cut failed on agreement and specificity. However, when consensus ratings were considered, the side hop met hypothesised values for all four outcomes. Furthermore, when an expert rater was used as the gold standard, rather than 3D motion analysis, both the drop jump and the side hop met hypothesised targets for all four outcomes.  Further analysis was performed to determine whether only one or two of these tasks would be needed to provide the same amount of information about an individual's degree of risk. Correlations between gold standard measures of reaching and knee abduction were calculated to determine whether there was a degree of interchangeability between tasks. Interestingly, a poor correlation (r=0.34) was found between side hop and side cut reaching values and only a 'fair' level of agreement was found between expert ratings for these two tasks (k=0.39). These results suggest that not all individuals who reach more on the side hop will reach more on the side cut, and vice versa. However, if one 'reaching' task had to be chosen over the other, the side hop would be preferable as it consistently outperformed the side cut on all aspects of agreement and validity.  There were weak, negative correlations of -0.30 and -0.33 between knee abduction values and lower limb reaching values. Kappa values for agreement between expert 87  ratings for drop jump and reaching tasks were negative (-0.16 and -0.06), representing worse than chance agreement. These results suggests that individuals who land with greater knee abduction angles on the drop jump do not necessarily reach more on the side hop and the side cut. Because knee abduction may be more prevalent in some individuals and reaching more prevalent in others, we would ultimately recommend that both elements need to be included in a comprehensive screening program.  4.5 Limitations A number of possible limitations may have affected our findings. In this study, only the biomechanics of the right lower limb were examined. This decision was based on evidence to suggest that the dominant leg displays more high risk neuromuscular characteristics than the non-dominant leg. In a study by Ford et al, side-to-side comparison of the lower limbs of high school basketball players during a drop jump landing showed significantly greater maximum valgus angles on the dominant side (27.6° ± 2.2°) compared to the non-dominant side (12.5° ± 2.80).24 This significant difference was observed in female athletes but not in males. A significant difference was also observed by Hewett et al in the maximum knee valgus angle between dominant and non-dominant limbs on a drop jump landing. 40 The difference was observed in female participants in the late or post pubertal stage but not in those in the early pubertal or pubertal stages. Admittedly though, when our method is eventually put to practical use in screening for injury risk, it may be important for coaches and clinicians to examine both lower limbs. Future research, requiring raters to examine both lower limbs, is warranted.  88  In a real-life situation, clinicians and coaches would examine participants 'live' rather than from video. For this study however, each participant took two hours to test in the lab and therefore it was not feasible to assemble the three raters to screen each new participant 'live'. While we did try to make conditions as 'real' as possible by not permitting the raters to pause, slow down or rewind the footage, the use of video, which is a two dimensional format, may have limited the validity of results because our 'gold standard' measure was three dimensional. Also, while the video footage was of a high quality, the camera angle, the smaller screen and the framing of the participant may have made the raters' jobs more difficult. Future studies, where raters evaluate athletes on the field, may reveal different results.  The study sample was selectively recruited in order to target the population most at risk of ACL injury. 284 However, if observational risk screening is ultimately to be implemented in the general athletic population it must first be tested with a more heterogeneous sample. In order to confirm the robustness of this screening method beyond the elite athletes tested in this study, further research needs to be conducted with a larger sample of randomly-selected athletes.  In this study, the three raters were all physiotherapists. Physiotherapists were chosen owing to their training and expertise in movement analysis and their experience in detecting inter-participant differences. Physiotherapists who work with sporting teams are well placed to perform risk screening. Ultimately however, it is likely that coaches and trainers would play a major role in performing risk screening with their teams. An important next step in this area of research is evaluating whether, with adequate 89  training, coaches and less experienced clinicians could obtain similar levels of agreement and validity.  90  5 CONCLUSION Until now, there have been no scientifically-tested methods to screen for ACL injury risk in the clinic or on the field. We developed practical and user-friendly observational risk screening guidelines to enable evaluation of neuromuscular characteristics associated with ACL injury. Our guidelines focussed on two common neuromuscular risk factors: increased knee abduction and lower limb reaching.  The decision of whether to adopt a new screening test must be based on its reliability, accuracy, ease of administration, cost of administration and its impact on patients.  58  In  this study, the only task which met all of these criteria was the side hop task, which was used to screen for lower limb reaching. With this task, raters met hypothesised intra and interrater agreement values and achieved a high level of sensitivity.  The unanticipated side cut task was also used to screen for lower limb reaching. This task had excellent 'ecological validity', replicating a manoeuvre commonly involved in ACL injury mechanisms. Unfortunately, it fared the worst of any task, failing to reach the majority of hypothesised agreement and validity levels. The side cut screening guidelines would therefore not be recommended for inclusion in a risk screening program at this stage.  As an ACL risk factor, the phenomenon of reaching has not been well researched, and the hypothesised effects of reaching on the ACL have not yet been proven. While this  91  characteristic has certainly been observed on video analysis of real-life ACL injuries, we do not yet know whether its presence may lead to future ACL injuries or how much reaching is too much. Further investigation into the biomechanics and prevalence of this risk factor is needed.  The drop jump task was used to screen for dynamic knee abduction. Raters exceeded hypothesised agreement values for this task but failed to reach targeted validity levels. Validity was negatively impacted by the seemingly poor equivalency between the 2D screening guidelines and the gold standard, 3D motion analysis. Validity was improved with the use of an expert rater as the gold standard. However, it must be acknowledged that the expert's ratings may or may not be an accurate reflection of ACL injury risk.  It is well accepted that a combination of factors, both intrinsic and extrinsic will combine to produce the perfect conditions for injury occurrence.106 Therefore, researchers, clinicians and coaches should avoid relying on a single variable or outcome measure when screening for ACL injury risk. This study has revealed two key components for inclusion in a neuromuscular risk screening program — dynamic knee abduction and lower limb reaching. However, it is important to remember that the intent is that they be used in conjunction with other risk indicators as part of a comprehensive screening package. Providing there is adequate reliability and validity, additional neuromuscular components such as reduced knee extension in loading, quadriceps dominance and limb dominance could also be included.  92  Ultimately, we look forward to a time when ACL risk screening programs are routinely implemented prior to involvement in all high risk sports. There is strong evidence to suggest that if detected, high risk neuromuscular characteristics can be improved with training. With accurate identification of athletes at increased risk of injury, prevention strategies can be targeted towards those most in need. Long term prospective studies may one day reveal a reduction in injury rate for high risk populations as a result of risk screening and targeted interventions.  93  6 BIBLIOGRAPHY 1. Anderson DV. Knee anatomy and mechanisms of injury. In: Sallis, R.E. and Massamino, F. (Eds.), Essentials of Sport Medicine. St. Louis: Mosby-Year Book; 1996, p.403-411.  2. Arendt E, Dick R. Knee injury patterns among men and women in collegiate basketball and soccer. NCAA data and review of literature. Am J Sports Med. 1995;23:694-701. 3. Blair SN, Brodney S. Effects of physical inactivity and obesity on morbidity and mortality: Current evidence and research issues. Med Sci Sports Exerc. 1999;31:S64662. 4. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307-310. 5. Boden BP, Dean GS, Feagin JA,Jr., Garrett WE,Jr. Mechanisms of anterior cruciate ligament injury. Orthopedics. 2000;23:573-8. 6. Bonci CM. Assessment and evaluation of predisposing factors to anterior cruciate ligament injury. J Athl Train. 1999;34:155-164. 7. Buehler-Yund C. A longitudinal study of injury rates and risk factors in 5 to 12 year old soccer players. In: Environmental Health. Cincinnati, Ohio: University of Cincinnati; 1999:161.  8. Bundy DG, Feudtner C. Preparticipation physical evaluations for high school athletes: Time for a new game plan. Ambul Pediatr. 2004;4:260-263. 94  9. Chmielewski TL, Hodges MJ, Horodyski M, Bishop MD, Conrad BP, Tillman SM. Investigation of clinician agreement in evaluating movement quality during unilateral lower extremity functional tasks: A comparison of 2 rating methods. J Orthop Sports Phys Ther. 2007;37:122-129.  10. Claiborne TL, Armstrong CW, Gandhi V, Pincivero DM. Relationship between hip and knee strength and knee valgus during a single leg squat. J App! Biomech. 2006;22:41-50.  11. Daniel DM, Stone ML, Dobson BE, Fithian DC, Rossman DJ, Kaufman KR. Fate of the ACL-injured patient. A prospective outcome study. Am J Sports Med. 1994;22:632644. 12. Davidson M. The interpretation of diagnostic test: A primer for physiotherapists. Aust J Physiother. 2002;48:227-232.  13. de Loes M, Dahlstedt LJ, Thomee R. A 7-year study on risks and costs of knee injuries in male and female youth participants in 12 sports. Scand J Med Sci Sports. 2000;10:90-97.  14. Dekker J, Tola P, Aufdemkampe G, Winckers M. Negative affect, pain and disability in osteoarthritis patients: The mediating role of muscle weakness. Behav Res Ther. 1993;31:203-206.  15. Dick R, Putukian M, Agel J, Evans TA, Marshall SW. Descriptive epidemiology of collegiate women's soccer injuries: National collegiate athletic association injury surveillance system, 1988-1989 through 2002-2003. J Athl Train. 2007;42:278-285.  95  16. Donner A, Eliasziw M. A goodness-of-fit approach to inference procedures for the kappa statistic: Confidence interval construction, significance testing and sample size estimation. Statistics in Medicine. 1992;11:1511-1519. 17. Dugan SA. Sports-related knee injuries in female athletes: What gives? Am J Phys Med Rehabil. 2005;84:122-130.  18. Eastlack ME, Arvidson J, Snyder-Mackler L, Danoff JV, McGarvey CL. lnterrater reliability of videotaped observational gait-analysis assessments. Phys Ther. 1991;71:465-472.  19. Eckhoff DG, Brown AW, Kilcoyne RF, Stamm ER. Knee version associated with anterior knee pain. Clin Orthop Relat Res. 1997,(339):152-155. 20. Eng JJ, Winter DA. Kinetic analysis of the lower limbs during walking: What information can be gained from a three-dimensional model? J Biomech. 1995;28:753758.  21. Fagenbaum R, Darling WG. Jump landing strategies in male and female college athletes and the implications of such strategies for anterior cruciate ligament injury. Am J Sports Med. 2003;31:233-240.  22. Ferretti A, Papandrea P, Conteduca F, Mariani PP. Knee ligament injuries in volleyball players. Am J Sports Med. 1992;20:203-207. 23. Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed. New York: John Wiley and Sons, Inc; 1981.  96  24. Ford KR, Myer GD, Hewett TE. Valgus knee motion during landing in high school female and male basketball players. Med Sci Sports Exerc. 2003;35:1745-1750. 25. Ford KR, Myer GD, Smith RL, Vianello RM, Seiwert SL, Hewett TE. A comparison of dynamic coronal plane excursion between matched male and female athletes when performing single leg landings. Clin Biomech (Bristol, Avon). 2006;21:33-40. 26. Ford KR, Myer GD, Toms HE, Hewett TE. Gender differences in the kinematics of unanticipated cutting in young athletes. Med Sci Sports Exerc. 2005;37:124-129. 27. Fuller CW, Ekstrand J, Junge A, et al. Consensus statement on injury definitions and data collection procedures in studies of football (soccer) injuries. Scand J Med Sci Sports. 2006;16:83-92.  28. Fung DT, Zhang LQ. Modeling of ACL impingement against the intercondylar notch. Clin Biomech. 2003;18:933-941.  29. Garrick JG. Preparticipation orthopedic screening evaluation. Clin J Sport Med. 2004;14:123-126. 30. Godshall RW. The predictability of athletic injuries: An eight-year study. J Sports Med. 1975;3:50-54.  31. Gomez JE, Landry GL, Bernhardt DT. Critical evaluation of the 2-minute orthopedic screening examination. Sports Med. 1993;147:1109-1113. 32. Grana WA, Moretz JA. Ligamentous laxity in secondary school athletes. JAMA. 1978;240:1975-1976.  97  33. Gray J, Taunton JE, McKenzie DC, Clement DB, McConkey JP, Davidson RG. A survey of injuries to the anterior cruciate ligament of the knee in female basketball players. Int J Sports Med. 1985 Dec;6:314-316. 34. Griffin LY, Agel J, Albohm MJ, et al. Noncontact anterior cruciate ligament injuries: Risk factors and prevention strategies. J Am Acad Orthop Surg. 2000;8:141-150. 35. Hartwick M, Meeuwisse W, Vandertuin J, Maitland M. Knee pain in the ACLdeficient osteoarthritic knee and its relationship to quality of life. Physiother Res Int. 2003;8:83-92. 36. Heidt RS,Jr, Sweeterman LM, Carlonas RL, Traub JA, Tekulve FX. Avoidance of soccer injuries with preseason conditioning. Am J Sports Med. 2000;28:659-662. 37. Hewett TE, Myer GD, Ford KR, Slauterbeck JR. Preparticipation physical examination using a box drop vertical jump test in young athletes. Clin J Sport Med. 2006;16:298-304.  38. Hewett TE, Ford KR, Myer GD. Anterior cruciate ligament injuries in female athletes: Part 2, a meta-analysis of neuromuscular interventions aimed at injury prevention. Am J Sports Med. 2006;34:490-498. 39. Hewett TE, Lindenfeld TN, Riccobene JV, Noyes FR. The effect of neuromuscular training on the incidence of knee injury in female athletes. A prospective study. Am J Sports Med. 1999;27:699-706. 40. Hewett TE, Myer GD, Ford KR. Decrease in neuromuscular control about the knee with maturation in female athletes. J Bone Joint Surg Am. 2004;86-A:1601-1608.  98  41. Hewett TE, Myer GD, Ford KR, et al. Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes: A prospective study. Am J Sports Med. 2005;33:492-501. 42. Hewett TE, Paterno MV, Myer GD. Strategies for enhancing proprioception and neuromuscular control of the knee. Clin Orthop Re/at Res. 2002;402:76-94. 43. Hickey BW, Milosavljevic S, Bell ML, Milburn PD. Accuracy and reliability of observational motion analysis in identifying shoulder symptoms. Man Ther. 2006.  44. Huston LJ, Wojtys EM. Neuromuscular performance characteristics in elite female athletes. Am J Sports Med. 1996;24:427-436.  45. Jacobs C, Mattacola C. Sex differences in eccentric hip-abductor strength and kneejoint kinematics when landing from a jump. Journal of Sport Rehabilitation. 2005;14:346355.  46. Jacobs CA, Uhl TL, Mattacola CG, Shapiro R, Rayens WS. Hip abductor function and lower extremity landing kinematics: Sex differences. J Athl Train. 2007;42:76-83.  47. Jebb SA, Moore MS. Contribution of a sedentary lifestyle and inactivity to the etiology of overweight and obesity: Current evidence and research issues. Med Sci Sports Exerc. 1999;31:S534-41.  48. Jian Y, Dinter DA, Ishac MG, Gilchrist L. Trajectory of the body COG and COP during initiation and termination of gait. Gait Posture. 1993;1:9-22.  49. Johnson RJ, Beynnon BD, Nichols CE, Renstrom PA. The treatment of injuries of the anterior cruciate ligament. J Bone Joint Surg Am. 1992;74:140-151.  99  50. Kalenak A, Morehouse CA. Knee stability and knee ligament injuries. JAMA. 1975;234:1143-1145. 51. Kernozek TW, Torry MR, VAN Hoof H, Cowley H, Tanner S. Gender differences in frontal and sagittal plane biomechanics during drop landings. Med Sci Sports Exerc. 2005;37:1003-12; discussion 1013. 52. Kibler WB, Chandler TJ, Uhl T, Maddux RE. A musculoskeletal approach to the preparticipation physical examination, preventing injury and improving performance. Am J Sports Med. 1989;17:525-531.  53. Knapik JJ, Bauman CL, Jones BH, Harris JM, Vaughan L. Preseason strength and flexibility imbalances associated with athletic injuries in female collegiate athletes. Am J Sports Med. 1991;19:76-81.  54. Knudson D. Validity and reliability of visual ratings of the vertical jump. Percept Mot Skills. 1999;89:642-648.  55. Krosshaug T, Nakamae A, Boden B, et al. Estimating 3D joint kinematics from video sequences of running and cutting maneuvers--assessing the accuracy of simple visual inspection. Gait Posture. 2007;26:378-385. 56. Krosshaug T, Slauterbeck JR, Engebretsen L, Bahr R. Biomechanical analysis of anterior cruciate ligament injury mechanisms: Three-dimensional motion reconstruction from video sequences. Scand J Med Sci Sports. 2007;17:508-519. 57. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159-174.  100  58. Lang TA, Secic M. How to Report Statistics in Medicine: Annotated Guidelines for Authors, Editors, and Reviewers. Philadelphia: American College of Physicians; 1997.  59. Liu SH, Al-Shaikh RA, Panossian V, Finerman GA, Lane JM. Estrogen affects the cellular metabolism of the anterior cruciate ligament. A potential explanation for female athletic injury. Am J Sports Med. 1997;25:704-709. 60. Loudon JK, Jenkins W, Loudon KL. The relationship between static posture and ACL injury in female athletes. J Orthop Sports Phys Ther. 1996;24:91-97. 61. Magee D. Orthopedic Physical Assessment. 4th ed. Philadelphia: Saunders; 2002. 62. Malinzak RA, Colby SM, Kirkendall DT, Yu B, Garrett WE. A comparison of knee joint motion patterns between men and women in selected athletic tasks. Clin Biomech (Bristol, Avon). 2001;16:438-45.  63. Mandelbaum BR, Silvers HJ, Watanabe DS, et al. Effectiveness of a neuromuscular and proprioceptive training program in preventing anterior cruciate ligament injuries in female athletes: 2-year follow-up. Am J Sports Med. 2005;33:1003-1010.  64. Markolf KL, Burchfield DM, Shapiro MM, Shepard MF, Finerman GA, Slauterbeck JL. Combined knee loading states that generate high anterior cruciate ligament forces. J Orthop Res. 1995;13:930-935. 65. McLean SG, Huang X, Su A, Van Den Bogert AJ. Sagittal plane biomechanics cannot injure the ACL during sidestep cutting. Clin Biomech (Bristol, Avon). 2004;19:828-838.  101  66. McLean SG, Huang X, van den Bogert AJ. Association between lower extremity posture at contact and peak knee valgus moment during sidestepping: Implications for ACL injury. Clin Biomech (Bristol, Avon). 2005;20:863-870. 67. McLean SG, Lipfert SW, van den Bogert AJ. Effect of gender and defensive opponent on the biomechanics of sidestep cutting. Med Sci Sports Exerc. 2004;36:1008-16. 68. McLean SG, Neal RJ, Myers PT, Walters MR. Knee joint kinematics during the sidestep cutting maneuver: Potential for injury in women. Med Sci Sports Exerc. 1999;31:959-68. 69. McLean SG, Walker K, Ford KR, Myer GD, Hewett TE, van den Bogert AJ. Evaluation of a two dimensional analysis method as a screening and evaluation tool for anterior cruciate ligament injury. Br J Sports Med. 2005;39:355-362. 70. McLean SG, Walker KB, van den Bogert AJ. Effect of gender on lower extremity kinematics during rapid direction changes: An integrated analysis of three sports movements. J Sci Med Sport. 2005;8:411-422. 71. Moore KL, Agur AMR. Essential Clinical Anatomy. 3rd ed. Baltimore: Lippincott Williams and Wilkins; 2007.  72. Myer GD, Ford KR, Brent JL, Hewett TE. Differential neuromuscular training effects on ACL injury risk factors in"high-risk" versus "low-risk" athletes. BMC Musculoskelet Disord. 2007;8:39.  102  73. Myklebust G, Engebretsen L, Braekken IH, Skjolberg A, Olsen OE, Bahr R. Prevention of anterior cruciate ligament injuries in female team handball players: A prospective intervention study over three seasons. Clin J Sport Med. 2003;13:71-78. 74. Neumann DA. Kinesiology of the Musculoskeletal System: Foundations for Physical Rehabilitation St. Louis: Mosby; 2002.  75. Norkin CC, White DJ. Measurement of Joint Motion: A Guide to Goniometty. Philadelphia: F.A. Davis Company; 1985.  76. Noyes FR, Barber SD, Mangine RE. Abnormal lower limb symmetry determined by function hop tests after anterior cruciate ligament rupture. Am J Sport Med. 1991;19:513-518. 77. Noyes FR, Barber-Westin SD, Fleckenstein C, Walsh C, West J. The drop-jump screening test: Difference in lower limb control by gender and effect of neuromuscular training in female athletes. Am J Sports Med. 2005;33:197-207.  78. Olsen OE, Myklebust G, Engebretsen L, Bahr R. Injury mechanisms for anterior cruciate ligament injuries in team handball: A systematic video analysis. Am J Sports Med. 2004;32:1002-12.  79. Orchard J, Seward H, McGivern J, Hood S. Intrinsic and extrinsic risk factors for anterior cruciate ligament injury in Australian footballers. Am J Sports Med. 2001;29:196-200. 80. Osterlind SJ. Modern Meaurement: Theory, Principles and Applications of Mental Appraisal. New Jersey: Pearson; 2006.  103  81. Patrick K, Norman GJ, Calfas KJ, et al. Diet, physical activity, and sedentary behaviors as risk factors for overweight in adolescence. Arch Pediatr Adolesc Med. 2004;158:385-390. 82. Petersen W, Braun C, Bock W, et al. A controlled prospective case control study of a prevention training program in female team handball players: The German experience. Arch Orthop Trauma Surg. 2005;125:614-621. 83. Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. 2nd ed. New Jersey: Prentice Hall Health; 2000.  84. Powell JW, Barber-Foss KD. Sex-related injury patterns among selected high school sports. Am J Sports Med. 2000;28:385-391. 85. Quatman CE, Ford KR, Myer GD, Hewett TE. Maturation leads to gender differences in landing force and vertical jump performance: A longitudinal study. Am J Sports Med. 2006;34:806-813.  86. Rankin G, Stokes M. Reliability of assessment tools in rehabilitation: An illustration of appropriate statistical analyses. Clin Rehabil. 1998;12:187-199. 87. Reikeras 0. Patellofemoral characteristics in patients with increased femoral anteversion. Skeletal Radio!. 1992;21:311-313. 88. Reinschmidt C, van den Bogert AJ, Nigg BM, Lundberg A, Murphy N. Effect of skin movement on the analysis of skeletal knee joint motion during running. J Biomech. 1997;30:729-732.  104  89. Roos H, Lindberg H, Gardsell P, Lohmander LS, Wingstrand H. The prevalence of gonarthrosis and its relation to meniscectomy in former soccer players. Am J Sports Med. 1994;22:219-222.  90. RumbaII JS, Lebrun CM. Preparticipation physical examination: Selected issues for the female athlete. Clin J Sport Med. 2004;14:153-160. 91. Russell KA, Palmieri RM, Zinder SM, Ingersoll CD. Sex differences in valgus knee angle during a single-leg drop jump. J Athl Train. 2006;41:166-171.  92. Segawa H, Omori G, Koga Y. Long-term results of non-operative treatment of anterior cruciate ligament injury. Knee. 2001;8:5-11. 93. Sell TC, Ferris CM, Abt JP, et al. The effect of direction and reaction on the neuromuscular and biomechanical characteristics of the knee during tasks that simulate the noncontact anterior cruciate ligament injury mechanism. Am J Sports Med. 2006;34:43-54. 94. Shambaugh JP, Klein A, Herbert JH. Structural measures as predictors of injury basketball players. Med Sci Sports Exerc. 1991;23:522-527. 95. Shultz SJ, Houglum P, Perrin DH. Examination of Musculoskeletal Injuries. 2nd ed. Champaign, Illinois: Human Kinetics; 2005.  96. Sigward SM, Powers CM. Loading characteristics of females exhibiting excessive valgus moments during cutting. Clin Biomech (Bristol, Avon). 2007;22:827-833. 97. Silbey MB, Fu FH. Knee Injuries. In: Fu, F.H. and Stone, D.A. (Eds.), Sports Injuries. Mechanisms, Prevention, Treatment. Baltimore: Williams and Wilkins; 1994, p. 949-976  105  98. Sim J, Wright CC. The kappa statistic in reliability studies: Use, interpretation, and sample size requirements. Phys Ther. 2005;85:257-268. 99. Slauterbeck JR, Fuzie SF, Smith MP, et al. The menstrual cycle, sex hormones, and anterior cruciate ligament injury. J Athl Train. 2002;37:275-278. 100. Soderman K, Werner S, Pietila T, Engstrom B, Alfredson H. Balance board training: Prevention of traumatic injuries of the lower extremities in female soccer players. A prospective randomized intervention study. Knee Surg Sports Traumatol Arthrosc. 2000;8:356-363.  101. Somers DL, Hanson JA, Kedzierski CM, Nestor KL, Quinlivan KY. The influence of experience on the reliability of goniometric and visual measurement of forefoot position. J Orthop Sports Phys Ther. 1997;25:192-202.  102. Streiner DL, Norman GR. Health Measurement Scales. A Practical Guide to their Development and use. 3rd ed. New York: Oxford University Press Inc.; 2003.  103. The National Collegiate Athletic Association. NCAA sports sponsorship and participation rates report: 1981-82 to 2005-06. Indianapolis:The National Collegiate Athletic Association; 2007.  104. Thijs Y, Van Tiggelen D, Willems T, De Clercq D, Witvrouw E. Relationship between hip strength and frontal plane posture of the knee during a forward lunge. Br J Sports Med. 2007.  105. Tooth LR, Ottenbacher KJ. The kappa statistic in rehabilitation research: An examination. Arch Phys Med Rehabil. 2004;85:1371-1376.  106  106. Uhorchak JM, Scoville CR, Williams GN, Arciero RA, St Pierre P, Taylor DC. Risk factors associated with noncontact injury of the anterior cruciate ligament: A prospective four-year evaluation of 859 west point cadets. Am J Sports Med. 2003;31:831-842. 107. von Porat A, Roos EM, Roos H. High prevalence of osteoarthritis 14 years after an anterior cruciate ligament tear in male soccer players: A study of radiographic and patient relevant outcomes. Ann Rheum Dis. 2004;63:269-273. 108. Walden M, Hagglund M, Ekstrand J. High risk of new knee injury in elite footballers with previous anterior cruciate ligament injury. Br J Sports Med. 2006;40:158-62; discussion 158-62.  109. Wojtys EM, Huston LJ, Boynton MD, Spindler KP, Lindenfeld TN. The effect of the menstrual cycle on anterior cruciate ligament injuries in women as determined by hormone levels. Am J Sports Med. 2002;30:182-188. 110. Woltring HJ. A FORTRAN package for generalized cross-sectional validatory spline smoothing and differentiation. Adv Eng Software. 1986;8:104-113. 111. Wright RW, Dunn WR, Amendola A, et al. Risk of tearing the intact anterior cruciate ligament in the contralateral knee and rupturing the anterior cruciate ligament graft during the first 2 years after anterior cruciate ligament reconstruction: A prospective MOON cohort study. Am J Sports Med. 2007;35:1131-1134. 112. Yeadon MR, Morlock M. The appropriate use of regression equations for the estimation of segmental inertia parameters. J Biomech. 1989;22:683-689. 113. Yu B, Lin CF, Garrett WE. Lower extremity biomechanics during the landing of a stop-jump task. Clin Biomech (Bristol, Avon). 2006;21:297-305. 107  114. Yu B, McClure SB, Onate JA, Guskiewicz KM, Kirkendall DT, Garrett WE. Age and gender effects on lower extremity kinematics of youth soccer players in a stop-jump task. Am J Sports Med. 2005;33:1356-1364. 115. Zazulak BT, Hewett TE, Reeves NP, Goldberg B, Cholewicki J. Deficits in neuromuscular control of the trunk predict knee injury risk: A prospective biomechanical-epidemiologic study. Am J Sports Med. 2007;35:1123-1130. 116. Zazulak BT, Hewett TE, Reeves NP, Goldberg B, Cholewicki J. The effects of core proprioception on knee injury: A prospective biomechanical-epidemiological study. Am J Sports Med. 2007.  108  7 APPENDICES 7.1 Appendix I. Consent Form THE t N VERSITA OF BRI FISH C 01,t BI A  UBC 1110.1111 OM*  /MI •••■ • ormiti  School of Rehabilitation Sciences Faculty of Medicine T325-2211 Wesbrook Mall Vancouver, British Columbia V6T 2B5 Phone: 604.822.7392 Fax: 604.822.7624 Web:  Consent Form — Subjects Visual Screening for High-Risk Landing Strategies in Female Soccer Players Principal Investigator: Dr. William C. Miller, School of Rehabilitation Sciences, UBC Co-Investigator(s): Ms. Christina Ekegren (MSc candidate), Mr. Richard Celebrini (PhD candidate), Drs. Janice Eng and Donna Maclntyre, School of Rehabilitation Sciences, UBC. Purpose: Soccer is a great sport but over the last few years an increasing number of young women are sustaining serious knee injuries. We are trying to see whether we can predict who might be at risk of getting injured. We are conducting a research project to develop and assess a new screening tool to detect the movement patterns thought to contribute to these serious knee injuries. You are invited to participate in this study because you are a healthy, 13-17 year old female soccer player. The information contained in this sheet will provide you with more details about the study so that you can decide whether you wish to participate. This study is being funded with a grant awarded by the BC Medical Services Foundation. Study Procedures: If you choose to participate in this study, you will be contacted to further explain the details of the study, answer any questions you may have, ensure that you qualify for the study, and to set up an initial testing session at the GF Strong Rehabilitation Center. The testing will consist of a 10-minute warm-up on a stationary cycle, followed by reflective markers and electrodes attached by tape to the leg, hip, and pelvis of your dominant side. Electrodes and markers will be applied by Ms. Ekegren. You will then be asked to perform 3 movements. First, you will stand in an athletic ready position and react as quickly as possible to a set of lights by planting your dominant leg on a platform and then kicking a  109  soccer ball to the left or right. Next, you will jump off a 30 cm box onto the platform. Finally you will perform a hop to the side. You will do all three of these movements 7 times after practicing several times. These movements will be assessed for the position and the forces on your knee joint and the activity of your muscles through video and electrical monitoring instruments. Three physiotherapists will view the videos of your performance and assess the way you move. You will not be identifiable from the video as it will be taken from the waist down.  Exclusion Criteria: You will not be eligible to participate in this study if you 1) have had a severe back or leg injury in the past, 2) have had any injury during the 6 weeks prior to the study, or 3) are presently using a supplemental exercise based training program such as a gym workout. Risk Section: One of the movements that you will be tested on is a quick reaction, cutting maneuver that is associated with a rare chance of risk (less than 2%). It is similar to movements you would perform during soccer training and playing. There have been no stated side effects of such testing in any of the similar studies published which have tested over 200 individuals. You will complete a warm-up and perform several practice trials to minimize any risk of injury. Signing this consent form in no way limits your legal rights against the sponsor, investigators, or anyone else. There will be no direct clinical benefit from participating in this study. However, at the end of the study you will be given the opportunity to receive the results of the study directly and feedback on your movement strategies. As well, any knowledge gained from the study will be shared with the soccer community including players, parents, and coaches. Stipend: You will be given a stipend of $20.00 to cover transportation, parking, and for your participation in the lab testing sessions. Confidentiality: Your identity will be kept confidential. All documents will be identified only by code number and kept in a locked filing cabinet. Your confidentiality will be respected. No information that discloses your identity will be released or published without your specific consent to the disclosure. However, research records and medical records identifying you may be inspected in the presence of the Investigator or his or her designate by representatives of Health Canada, and the UBC Research Ethics Board for the purpose of monitoring the research. However, no records which identify you by name or initials will be allowed to leave the Investigators' offices. The only people having access to this information will be the investigators mentioned above and the research assistants. Contact for information about the study: If you have any questions or desire further information with respect to this study, you may contact Dr. Bill Miller or Christina Ekegren at 604 734 1313 (ext. 2118). Contact for concerns about the rights of research subjects: If you have any concerns about your treatment or rights as a research subject, you may contact the Research Subject Information Line in the UBC Office of Research Services at 604-822-8598. Disclosure Regarding Rights of Subject to Withdraw from the Research Your participation in this study is entirely voluntary and you may refuse to participate or withdraw from the study at any time without jeopardy to your place within the team. You do not have to provide any reasons for your decision. Data collected up to the point of your withdrawal from the study must be kept for data 110  analysis purposes under strict provisions of confidentiality. Unless specified otherwise you may be contacted in the future for related studies. At that time you can refuse to participate and your name will be removed from future correspondence. If you decide to participate in future studies, you will be asked to sign another consent form specific to that study.  111  Consent: For parent/guardian if subject less than 19 years of age Your signature below indicates that you have received a signed and dated copy of this consent form for your own records. Your signature does not waive any of your/your daughter's legal rights. Your signature indicates that you consent for your daughter to participate in this study Please return this portion of the consent form in the self addressed envelope and keep the rest of the form for your records.  I consent to my daughter's participation in this study.  Name of Parent/guardian^(Please Print)  Name of Witness^(Please Print)  Signature of Parent/guardian^Date^Signature of witness^Date  Name of Investigator^(Please Print)  Name of translator^(Please print)  Signature of Investigator^Date^Signature of translator^Date  Language of translation  I wish to be contacted for future studies -^YES NO  112  Assent: For subjects less than 19 years of age I have had the opportunity to read this consent form, to ask questions about my participation in this research, and to discuss my participation with my parents. All my questions have been answered. I understand that I may withdraw from this research at any time, and that this will not interfere with the availability to me of other health care. I have received a signed and dated copy of this consent form. If I write and sign my name below, it means that I agree to be in the study.  Printed name of subject  Signature^  Date  113  7.2 Appendix II. Information Sheet  Name:^  Date of birth: (yyyy/mm/dd)  Address: ^Postal  code: ^  Home phone: ^  Cell: ^  E-mail: ^ Next of kin (eg. parent's name): ^ Relationship to you: ^ Phone no: daytime: ^  evening: ^  Leg dominance  R^L Which leg would you feel most comfortable kicking a ball with?^0^0 Injuries Have you had any injuries/conditions/surgeries that have interfered with sport?^  Yes No Unsure CI^0^0  For each injury/condition/surgery: Description (include side of injury)^Date^Any residual problems?  114  7.3 Appendix III. Limb Length and Circumference Measures  Measurements (cm):  Foot: h^Length:^Heel to toe P1^Perimeter: Min near ankle P2^Perimeter: Arch P3^Perimeter:^Ball Leg: h^Length:^Knee centre to ankle centre P1^Perimeter: Knee P2^Perimeter: Max P3^Perimeter: Min near ankle  115  7.4 Appendix IV. Postural Measurements61 1. Rearfoot eversion/inversion  a. In standing, mark Achilles insertion and a point 1cm below bisecting calcaneus b. Make 2 marks in lower 1/3 leg, bisecting calf c. Measure angle b/w 2 lines d. Lines should be parallel or between 2-8° varus 2. Navicular drop  a. In standing put subject in neutral talus position b. Find neutral talus by palpating head of talus medially and laterally. Ask subject to turn to left then right and find a position where talus does not bulge on either side. c. Measure distance from base of navicular to floor d. Allow subject to return to relaxed stance e. Measure distance from base navicular to floor f. If >10mm, excessive navicular drop 3. Knee varus/valgus  a. Align subject with patellae facing forward and knees and medial malleoli touching b. If knees touch and ankles do not measure between medial malleoli (>910cm is excessive valgus) c. If ankles touch and knees do not measure between knees (>4cm/2 fingers is excessive varus) 4. Knee hyperextension/flexion  a. Ask subject to fully extend knees b. Align goniometer along long axis of fibula and femur c. <0° or >5° is excessive extension/flexion  116  7.5 Appendix V. IRED Marker Positions  IRED no.  Digitised points  1  Head of 5th metatarsal  2  Dorsal foot (midpoint of metatarsals and ankle joint)  3  Lateral heel  4  Lateral malleolus  5  Mid-shank (anterior aspect of tibia, midpoint of ankle & knee)  6  Head of fibula  7  Middle of tibia  8  Lateral femoral condyle  9  Lower thigh  10  Greater trochanter  11  Medial femoral condyle  12  Middle of femur  1  Head of 2nd MTP  2  Medial malleolus  3  Medial tibial condyle  4  Medial femoral condyle  5  Right ASIS  6  Left ASIS  7  Right superior iliac crest  117  7.6 Appendix VI. Ethics Certificate of Approval The University of British Columbia Office of Research Services Clinical Research Ethics Board — Room 210, 828 West 10th Avenue, Vancouver, BC V5Z 1L8  LiBC  ETHICS CERTIFICATE OF EXPEDITED APPROVAL PRINCIPAL INVESTIGATOR:  INSTITUTION / DEPARTMENT:  UBC/Medicine, Faculty of/Rehabilitation Sciences  William C. Miller  UBC CREB NUMBER: H07 - 00248  INSTITUTION(S) WHERE RESEARCH WILL BE CARRIED OUT: Institution^  I^  Site  Vancouver Coastal Health (VCHRI/VCHA)^GF Strong Rehabilitation Centre Other locations where the research will be conducted:  N/A  CO - INVESTIGATOR(S):  Christina Ekegren Janice Eng Richard Celebrini Donna L. Maclntyre  SPONSORING AGENCIES: BC Medical Services Foundation - "The effectiveness of a novel warm-up in decreasing risk factors for anterior cruciate ligament injury in female youth soccer players"  PROJECT TITLE: Visual Screening for High-Risk Landing and Cutting Strategies in Female Soccer Players  THE CURRENT UBC CREB APPROVAL FOR THIS STUDY EXPIRES: March 5, 2008 The UBC Clinical Research Ethics Board Chair or Associate Chair, has reviewed the above described research  project, including associated documentation noted below, and finds the research project acceptable on ethical grounds for research involving human subjects and hereby grants approval.  DOCUMENTS INCLUDED IN THIS APPROVAL: Pocument Name^  APPROVAL DATE: 'Version  Protocol: Protocol^  Date^I  2  February 26, 2007  3  February 26, 2007  2  February 26, 2007  Consent Forms: Consent Form^  Assent Forms: Assent Form^  March 5, 2007  Questionnaire, Questionnaire Cover Letter, Tests: Data collection sheet^  1  January 31, 2007  2  February 26, 2007  Letter of Initial Contact: Letter of Initial Contact^  118  CERTIFICATION: In respect of clinical trials:  1. The membership of this Research Ethics Board complies with the membership requirements for Research Ethics Boards defined in Division 5 of the Food and Drug Regulations. 2. The Research Ethics Board carries out its functions in a manner consistent with Good Clinical Practices. 3. This Research Ethics Board has reviewed and approved the clinical trial protocol and informed consent form for the trial which is to be conducted by the qualified investigator named above at the specified clinical trial site. This approval and the views of this Research Ethics Board have been documented in writing. The documentation included for the above-named project has been reviewed by the UBC CREB, and the research study, as presented in the documentation, was found to be acceptable on ethical grounds for research involving human subjects and was approved by the UBC CREB. Approval of the Clinical Research Ethics Board by one of:  Dr. Caron Strahlendorf, Associate Chair  119  7.7 Appendix VII. Rating Form  DROP JUMPS^  ID#  171^Comments  Subject # 1  0 HR  0 LR  2  U HR  0 LR  3  U HR  U LR  4  U HR  U LR  5  U HR  U LR  6  U HR  U LR  7  0 HR  0 LR  8  CI HR  U LR  9  0 HR  U LR  10  U HR  0 LR  11  0 HR  0 LR  12  LI HR  U LR  13  O HR  0 LR  14  0 HR  U LR  15  U HR  U LR  16  U HR  0 LR  17  0 HR  U LR  18  LI HR  U LR  19  U HR  U LR  20  U HR  0 LR  21  U HR  LI LR  22  0 HR  U LR  23  El HR  0 LR  24  U HR  0 LR  25  U HR  U LR  26  LI HR  D LR  27  0 HR  U LR 120  28  0 HR  U LR  29  U HR  U LR  30  U HR  U LR  31  0 HR  0 LR  32  0 HR  El LR  33  U HR  U LR  34  0 HR  U LR  35  U HR  U LR  36  0 HR  U LR  37  U HR  U LR  38  0 HR  Li LR  39  U HR  U LR  40  U HR  0 LR  121  7.8 Appendix VIII. Standard Error of Measurement  SEM = s-N11- r  Where, s = sample standard deviation r = reliability coefficient (ICC)  122  7.9 Appendix IX. ICC 3 k Formulae  ICC3,k  BMS — EMS BMS  ^ = between subjects mean square (from ANOVA) Where,^BMS ^ EMS = error mean square  123  7.10 Appendix X. Bland and Altman Formulae  BRDiff  = Baseline score- retest score  BRMean  Baseline + retest =^  2  1BRDiff Mean Difference ^=^n  Limits of agreement = mean difference ± (2xSD(BRDiff))  124  7.11 Appendix XI. Kappa Scale of Agreement 57  0.21-0.40 = 'Fair' 0.41-0.60 = 'Moderate' 0.61-0.80 = 'Substantial' 0.81-1.00 = 'Almost perfect'  125  7.12 Appendix XII. Kappa: 2x2 Contingency Tables and Calculations  To calculate kappa, a 2x2 contingency table is formed. For intrarater agreement, individual subjects are put into box A, B, C or D depending on their ratings at different time points. For interrater agreement, individual subjects are put into box A, B, C or D depending on their ratings from different raters.  2x2 table for intrarater agreement  Rater 1  Time 1 High-risk^Low-risk  Total  Time 2^High-risk  A  B  B1 = A+B  Low-risk  C  D  B2 = C-FD  Total  Al = A+C^A2 = B+D  N  2x2 table for interrater agreement  Time 2  Rater 1 High-risk^Low-risk  Total  Rater 2^High-risk  A  B  B1 = A+B  Low-risk  C  D  B2 = C+D  Total  Al = A+C^A2 = B+D  N  126  Calculations for kappa23 p - e(x-)  -^ 1 - e(x)  Where, (A+ D)  e(K) =  Al\ ( B1)^A2)(B2\ \NI\N ^NA Ni  Calculations for multirater kappa23 P - Pe , 1-  pe  Where, 17. = mean 'Agreement' for all participants (n=40)  1= 1  Where, ' Agreement' =  - 1) E ny(no K(K - 1)  ny = number of raters who classified participant i in category j K = total number of raters (3) R = number of decision categories (2) = proportion of all classifications that fall within each decision category  127  7.13 Appendix XIII. Kappa Assumptions105  A study must meet the following assumptions in order to obtain valid kappa values: 1. Subjects and observations are independent of each other 2. Raters score subjects in an independent fashion 3. Rating categories are mutually exclusive and exhaustive  128  7.14 Appendix XIV. Validity: 2x2 Contingency Tables and Calculations  As for kappa, a 2x2 contingency table is formed. Individual subjects are put into box a, b, c or d depending on their ratings from observational screening and 3D motion analysis. Following this, sensitivity, specificity, positive predictive value and negative predictive value can be calculated.  2x2 contingency table  Results of 3D motion analysis  High-risk(?x°)^Low-risk (<x°) Results of^High-risk  Total  a  b  a+b  C  d  c+d  observational screening^Low-risk  Total  a + c^b + d  n  Calculations  Sensitivity =  PV+ =  a a+ c  a a+ b  Specificity =  d  b+ d  d c+ d  PV = ^ -  129  7.15 Appendix XV. Magnitude of Correlations83  Correlation Poor Moderate Good  Negative  Positive  >-0.50  <0.50  -0.50 to -0.75 -0.50 to -0.75 <-0.75  >0.75  130  7.16Appendix XVI. Prevalence of High Risk-Low Risk Physiotherapist Ratings*  HR Drop Jump LR HR Side Hop LR HR Side Cut LR  Rater 1  Rater 2  Rater 3  Time 1^Time 2  Time 1^Time 2  Time 1^Time 2  18  22  21  18  20  17  (0.45)  (0.55)  (0.53)  (0.45)  (0.50)  (0.43)  22  18  19  22  20  23  (0.55)  (0.45)  (0.47)  (0.55)  (0.50)  (0.57)  30  31  27  28  27  26  (0.75)  (0.78)  (0.68)  (0.70)  (0.68)  (0.65)  10  9  13  12  13  14  (0.25)  (0.22)  (0.32)  (0.30)  (0.32)  (0.35)  31  38  30  27  30  28  (0.78)  (0.95)  (0.75)  (0.68)  (0.75)  (0.70)  9  2  10  13  10  12  (0.22)  (0.05)  (0.25)  (0.32)  (0.25)  (0.30)  Abbreviations: HR, high risk; LR, low risk *The prevalence of all ratings is included in parentheses  131  7.17 Appendix XVII. Prevalence of 'Expert' and 'True' Ratings Expert Ratings^True Ratings  Prevalence^N^Prevalence Drop Jump  Side Hop  Side Cut  HR^21^0.575^15^0.375 LR^19^0.425^25^0.625  HR^24^0.60^22^0.55 LR^16^0.40^18^0.45  HR^22^0.55^16^0.40 LR^18^0.45^24^0.60  132  • •  7.18Appendix XVIII. ROC Curve for Drop Jump Task* 1.0 0.9 •5 0.8 -  c 0.7 0.6  -  Knee abduction,, motion  ce > 0.4  o o  a-  0.3 0.2 0.1 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0  False Positive Rate (1-Specificity)  *Diagonal line represents no discrimination  133  • ▪  7.19 Appendix XIX. ROC Curve for Side Hop Task*  1.0 ^  • •^ 1:19 •  Cut point = 0.29  > 0.8 • co^• • c 0.7 1 w : Cl)^• — 0.6 'vs 0.5 W  Reach]  > 0.4:III (S, 0.3 a. a) 0.2 2 i=" 0.1 0.0 ^ 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1  False Positive Rate (1-Specificity)  *Diagonal line represents no discrimination  134  7.20 Appendix )(X. ROC Curve for Side Cut Task*  0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1  False Positive Rate (1-Specificity)  *Diagonal line represents no discrimination  135  7.21 Appendix )0(1. Correlations Between Knee Abduction Motion and Static Postural Characteristics of Right Lower Limb Knee Abduction Pearson's r  Significance  Spearman's  Significance  (2 tailed)  rho  (2 tailed)  Knee Valgus  0.21  0.19  0.22  0.17  Knee Varus  -0.20  0.22  -0.27  0.10  Rearfoot Eversion  -0.06  0.72  -0.10  0.55  Navicular Drop  0.03  0.86  -0.01  0.95  Knee Hyperextension  -0.09  0.59  -0.07  0.68  136  7.22 Appendix XXII Frequency Distribution- Knee Abduction Motion  10  8•  6•  4•  2•  II  Std. Dev = 3.52 Mean = 9.8 N = 40.00  2.0^4.0^6.0^8.0^10.0^12.0^14.0^16.0 3.0^5.0^7.0^9.0^11.0^13.0^15.0^17.0  Knee Abduction Motion (degrees)  137  7.23Appendix XXIII Frequency Distribution- Side Hop Reaching  Std. Dev = .09 Mean = .295 N=40.00 .125^.175^.225^.275^.325^.375^.425 .150^.200^.250^.300^.350^.400^.450  Amount of Reach  138  7.24Appendix )0(IV Frequency Distribution- Side Cut Reaching  .200 .225 .250 .275 .300 .325 .350 .375 .400 .425  139  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items