Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics

JCLB-03998; No of Pages 7 Clinical Biomechanics xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Clinical Biomechanics journal homepage: www.elsevier.com/locate/clinbiomech

Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics Edward Nyman Jr. ⁎, Charles W. Armstrong Motion Analysis Laboratory, Department of Kinesiology, College of Health Sciences, The University of Toledo, Toledo, OH, USA

a r t i c l e

i n f o

Article history: Received 13 April 2015 Accepted 23 June 2015 Keywords: Anterior cruciate ligament Injury prevention Knee injury Kinect

a b s t r a c t Background: Although neuromuscular training featuring visual feedback may benefit modification of anterior cruciate ligament injury-risk linked knee kinematics, wide-spread clinical intervention has been limited to date. This study evaluated the effects of a Microsoft Kinect-based feedback system for modification of drop vertical jump knee kinematics traditionally consistent with predisposition to non-contact anterior cruciate ligament injury in female athletes. We hypothesized that a four-week feedback training protocol would increase peak knee flexion angle and frontal plane valgus-correlated knee separation distance during drop jump landing performance. Methods: Twenty-four female athletes were randomly divided equally into control or Kinect-based feedback groups. Subjects were pre-screened for peak knee flexion angle and minimum knee separation distance during drop landing and later performed twenty 31 cm drop landings three days per week for four weeks. The feedback group received Kinect-based visual feedback, while controls did not. Kinematics were re-assessed immediately following the end of the training period. Findings: The feedback group increased peak knee flexion and experienced a greater improvement in peak knee flexion. The feedback group improved normalized knee separation distance with greater improvement in post-training peak knee separation distance as compared with controls. Interpretation: Kinect-based feedback training significantly improved drop vertical jump knee kinematics associated with non-contact anterior cruciate ligament injury. The Kinect-based feedback approach demonstrates promise for mitigating non-contact anterior cruciate ligament injury predisposing knee biomechanics in female athletes within the clinical environment. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Each year, over 200,000 anterior cruciate ligament (ACL) injuries occur in the United States alone, with approximately one third involving female athletes (Hughes and Watkins, 2006; Chaudhari et al., 2007). With most incidences requiring reconstruction surgery at an average affiliated medical cost approaching $20,000 per case (Grindstaff et al., 2006; Sugimoto et al., 2012), the estimated aggregate healthcare cost is more than one billion US dollars per year (Hughes and Watkins, 2006; Kramer et al., 2007; Dallinga et al., 2012; Donnelly et al., 2012; Sugimoto et al., 2012). Nearly 70% of ACL injury cases are non-contact in nature (McNair et al., 1990; Agel et al., 2005), and females fall victim to this injury four to seven times more frequently than males (Ford et al., 2005a; Hewett et al., 2005a; Myer et al., 2005; Zazulak et al., 2006). Female athletes in soccer and gymnastics account for 4.53

⁎ Corresponding author at: Engineering Center for Orthopaedic Research Excellence (ECORE), Departments of Bioengineering and Orthopaedic Surgery, Colleges of Engineering and Medicine, MS 303, College of Engineering, The University of Toledo, Toledo, OH 43606, USA. E-mail address: [email protected] (E. Nyman).

and 4.23 knee injuries per 10,000 athletic exposures, respectively (Swenson et al., 2013). In 2013, these rates were the highest rates for female sports and were exceeded only by American football (Swenson et al., 2013) overall. It is estimated that over 50% of ACL-injured knees will develop osteoarthritis in as little as ten years (Hootman and Albohm, 2012). Thus, ACL injury is not only a serious issue for athletes, but has profound financial and quality of life implications for individuals across their lifespan. Most non-contact ACL injuries occur during sudden deceleration, rapid change of direction, or while landing from a jump (Paszkewicz et al., 2012). It is under these dynamic conditions, and in particular within the first 30 to 50 ms after initial ground contact (Koga et al., 2010), that the ACL appears to be most vulnerable to injury from a biomechanical perspective. Females are believed mitigate ground reaction force at the knees in a different manner than their male counterparts, particularly in the frontal and sagittal planes (Hewett et al., 2004, 2005b; Ford et al., 2005b; Hewett, 2008). The resultant dynamic knee valgus has been implicated as a major contributing factor to higher non-contact ACL injury rates in female athletes (Hewett et al., 2006a; Quatman and Hewett, 2009). This characteristic aberrant frontal plane motion is correlated with non-contact ACL injury, is readily identifiable,

http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018 0268-0033/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

2

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx

and is therefore useful as a clinical biomarker for injury predisposition. Decreased sagittal plane knee flexion has also been implicated in elevated risk of ACL injury (Hewett et al., 2004, 2005a; Quatman and Hewett, 2009; Quatman et al., 2014) as the knee is at increased biomechanical risk, from a physiological alignment and musculoskeletal joint protection standpoint, when loaded at reduced angles of flexion (DiStefano et al., 2011; Dowling et al., 2012). Thus, reduced knee flexion is an additional clinical biomechanical marker for ACL injury risk. As knee flexion angle and knee separation distance, during the first 50 ms following initial ground contact, have been implicated in increased predisposition to non-contact ACL injury in the female athlete, these bilateral metrics are therefore appropriate for tracking and guiding real-time feedback for motor learning and neuromuscular training purposes. Neuromuscular training has been explored extensively in efforts reduce non-contact ACL injury risk as the knee. Knee separation distance has also been targeted as it can provide a surrogate representation of relative knee valgus during drop vertical jump screening (Noyes et al., 2005). This implication is important as it enables knee separation distance to act as a surrogate for frontal plane knee valgus in single camera view tracking approaches. While ACL injury is a multi-factorial problem, a recent trend for ACL researchers and clinicians is to focus intervention efforts on the most readily modifiable factors (DiStefano et al., 2011) as these factors may be addressed non-invasively in the clinic. Neuromuscular control, lower extremity muscular strength, and capacity for proper mitigation of ground reaction forces are all therein deemed potentially modifiable. Recent research has shown that modification of these factors may reduce ACL injury rates (Noyes et al., 2005; Dowling et al., 2012). With established data linking aberrant neuromuscular control during drop vertical jump (DVJ) with elevated non-contact ACL injury risk (Zazulak et al., 2007), the need for accurate quantitative feedback systems that can target neuromuscular control deficits has become paramount. While feedback training has demonstrated success in modifying biomechanics consistent with risk (Dowling et al., 2012; Myer et al., 2013; Stroube et al., 2013), elaborate laboratory-based motion capture systems have traditionally been utilized for this purpose (Onate et al., 2005; Herman et al., 2009; Barrios et al., 2010). However, there are several factors that preclude the practical use of traditional systems in a clinic or field-based environment. These limitations include the high cost of the equipment, requisite additional technical training for the clinician, and the lack of portability of such a system. Logically, a feedback system that provides the benefits of contemporary laboratory-based biomechanics instrumentation, without the significant logistical limitations, would be ideal for field and clinical use, thus bringing intervention benefits to more individuals at risk of injury. In recent years, a number of intervention studies using feedback training (Dowling et al., 2012; Myer et al., 2013; Stroube et al., 2013) have been conducted wherein visual and verbal real-time or delayed feedback has contributed to demonstrated improvements in kinematics. In one such study, Stroube et al. (2013) investigated the effects of video feedback on kinematics during repeated tuck-jump training bouts and found that subjects who received such feedback were able to significantly reduce kinematic deficits. Additionally, Myer et al. (2013) reported similar findings supporting the benefit of augmented feedback on modification of landing mechanics consistent with noncontact ACL injury risk. Other methods have been successfully implemented for real-time and delayed feedback, such as that by Dowling et al. (2012) who utilized strategically placed inertial sensors on lower extremity segments to provide real-time feedback during a training protocol. Though it has been previously established that feedback may aid improvement of neuromuscular control related to knee function (Brindle et al., 2009; Barrios et al., 2010; Benjaminse et al., 2010), realtime kinematic feedback training has not yet been widely integrated at the clinical level. However, as recent innovations in hardware and software have made intuitive markerless tracking of human movement

feasible (Dutta, 2012; Mentiplay et al., 2013; Kiefer et al., 2015), the potential to provide real-time tracking of joint kinematics outside of a traditional motion analysis laboratory environment is now realizable (Wong, 2011; Borenstein, 2012). With the advent of the Microsoft Kinect (Microsoft Corporation, Redmond, WA, USA) depth camera system, PrimeSense (Tel Aviv, Israel) provided Microsoft with a platform capable of real-time capture and recording of human kinematics outside of a traditional motion analysis laboratory (Wong, 2011; Borenstein, 2012). The Kinect depth camera uses calibrated structured infrared light emission and detection to track limb segments within a defined and pre-calibrated physical capture volume (Borenstein, 2012; Obdralek, 2012). The small space requirements, ease of use, and low cost make a Kinect-based feedback training approach attractive for clinical biomechanics applications. In previous work by the authors, a Kinect-based knee kinematics tracking system for clinical ACL injury risk screening was developed and successfully validated against a traditional laboratory-based motion capture approach (Nyman, 2013). The purpose of this study was to evaluate the effects of Kinect-based real-time feedback on knee kinematics consistent with non-contact ACL injury risk during drop jump landing training in young female athletes. We hypothesized that a four-week real-time feedback training protocol would be effective at improving bilateral peak knee flexion angle (sagittal plane) and peak knee separation distance (frontal plane) during the first 50 ms following initial ground contact in drop jump landings. 2. Methods 2.1. Subjects A convenience sample of thirty-two female adolescent gymnasts were initially recruited from local competitive gymnastics programs, of which twenty-four were successfully enrolled and ultimately completed the study as approved by the University of Toledo Biomedical Institutional Review Board (IRB) and conducted in accordance with the measures outlined in the Declaration of Helsinki. All subjects assented to participate and a parent or legal guardian signed informed consent. Subjects were required to be free from recent lower extremity and spine musculoskeletal injury, have no history of orthopaedic surgery, no recent concussions, and intact bilateral ACLs. Each subject that met the inclusion criteria was randomly assigned to one of two groups: control (CTRL) or Kinect-based feedback (KBF). Mean subject age was: 15.0(1.6) years, mean height was 1.60(0.07) meters, mean body mass was 53.4(7.1) kg, and mean competitive gymnastics ability was level 7(1), on a scale of level 2 (lowest) through level 10 (highest) as established by the national governing body of the sport in the United States (USA Gymnastics, Indianapolis, IN, USA). Final group sizes were 12 CTRL and 12 KBF participants. 2.2. Instrumentation Custom software was developed within an open source framework to create a real-time user-intuitive Kinect-based visual and auditory feedback system. This was initiated with the intention of future opensource provision to interested clinicians in order to facilitate rapid deployment into clinical fields. The primary components of the KBF system included a commercially available infrared depth-camera (Microsoft Kinect, Microsoft Corporation, Redmond, WA, USA), open source drivers (NiTE, PrimeSense, Tel Aviv, Israel), open source programming (OpenNI.org), and graphical user interface software (Processing.org). The complete system was completed by housing all components, including laptop PC (Lenovo, Beijing, China) and 24-inch LED monitor (Samsung, Seoul, South Korea), on an industrial rolling cart (Fig. 1). In order to present real-time skeletal tracking and display on the viewing screen, raw three-dimensional coordinate data from the

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx

3

24” HD LED Monitor

Kinect Depth Camera

Windows Laptop PC

Rolling Metal Cabinet

Power Supply & Cables

Fig. 1. Kinect-based feedback (KBF) system for clinical application.

camera were filtered and processed by an imbedded software algorithm. The randomized decision forest algorithm constructed “skeleton” segments which were derived from a fusion of two-dimensional (x and y coordinate) pixel tracking with the third dimensional point value provided by the depth image value (z coordinate) and then overlaid onto a visual representation of the subject's silhouette as colored line segments on the viewing screen (Borenstein, 2012) (Fig. 2). Lower extremity segments were tracked and plotted in real-time, with minimal latency, at a rate of 30 Hz in all three dimensions and planes of movement (Wong, 2011; Borenstein, 2012). Frontal and sagittal angular rotations were calculated and logged in time series format. Relevant angles (knee flexion) and distance (intercondylar knee separation distance) were utilized to drive the primary visual and auditory feedback components, which consisted of on-screen graphical representation of lower extremity limb and joint segments that responded with nearly zero latency to subject movement within the prescribed capture volume. Frontal plane knee separation distance and sagittal

plane knee flexion angles provided values accurate to within 0.5°, consistent with the findings of our previous work (Nyman, 2013) as well as Schmitz et al. (2014). Subjects were initially recognized by the real-time tracking system by producing the system-specific subject calibration kinematic (“Psi”) pose (Fig. 2), briefly, until the system successfully recognized the subject, began tracking their movement, and displayed a dynamic line between the subject's knee joints that matched the subject's intercondylar knee separation distance. During dynamic activity, as knee separation distance increased or decreased, the length and color of the line changed accordingly. When this distance was less than the desired knee separation “threshold” (described below), the line color was displayed in red, and when the distance was equal to or greater than the threshold, the line color was displayed in green. The knee separation “threshold” was calculated for each subject in real-time by computing the distance between the subjects' left and right anterior superior iliac spine (inter-ASIS) locations and setting the intercondylar knee

Fig. 2. Graphical user interface screen & subject calibration pose.

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

4

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx

separation distance threshold to fifty percent of the inter-ASIS distance. This approach was influenced by previous threshold approaches in the literature such as that of Myer et al. (2013) who tracked maximum medial knee displacement as part of their feedback protocol. Additionally, the real-time quantitative value of the resultant knee separation distance vector was displayed as text, with values in millimeters, at the top of the screen in accordance with the color scheme previously detailed. Knee flexion angles were also calculated in real-time and displayed, in degrees, at the top of the screen according to a similar convention. This text was displayed in green when the knee flexion angle was greater than thirty degrees and in red when the value was less than thirty degrees. Additionally, an audible tone was emitted from the PC speakers when either of the tracked kinematics dropped below the prescribed thresholds. 2.3. Data collection Participants completed all testing and intervention training barefoot and wearing standard form-fitting gymnastics leotard uniform clothing. Subjects were familiarized to the 31 cm box drop vertical jump (DVJ) landing task via verbal explanation and expert demonstration, after which subjects had the opportunity to practice the movement five times. During testing sessions, subjects were all verbally instructed to “jump off the box, land on both feet, and rebound into a maximum effort vertical jump.” No visual or verbal feedback with regards to the DVJ landing technique was offered to subjects at the time of pre-test or post-test. Once the subject was comfortable performing the task, initial kinematic (pre-test) data were collected with a Kinect-based screening system positioned in front of the subject at a distance of 3 m and a height of 1 m relative to the jump landing area. This method of kinematic data collection has previously been validated for such purposes as demonstrated in our earlier work (Nyman et al., 2013) and by that of Schmitz et al. (2014) and Mentiplay et al. (2013). Peak knee flexion angle in the sagittal plane and minimum knee separation distance in the frontal plane were recorded simultaneously from a single camera view, using the Kinect system, for each of five independent trials for each subject. These data were then stored for later analysis and comparison. 2.4. Feedback intervention training program The four week intervention training period began immediately following the initial kinematic pre-test. All training sessions were provided in the athletes' training facility and utilized identical gymnasticsspecific landing surfaces (2″ carpet-bonded foam floor exercise matting) that were familiar and consistent with their regular sport activities. Subjects wore the identical type and fit of gymnastics leotard that was utilized during the testing sessions. For the Kinect-based feedback (KBF) group, each subject performed a series of twenty 31 cm drop landings using the Kinect-based real-time feedback system three days per week for four consecutive weeks. Subjects were instructed to complete each jump separately, using the real-time Kinect-derived images presented to them on the monitor, placed at average subject eye-level (1.25 m from floor) and 2 m in front of the landing area, to guide their landing mechanics in accordance with the intervention protocol. The real-time display presented subjects with qualitative and quantitative information, in simple graphical format, using colored text, colored lines, and the auditory tone described in the previous section (Fig. 3). A green line connecting the knees indicated proper spacing in the frontal plane, while a red line indicated aberrant frontal plane knee mechanics. KBF subjects were instructed to use the feedback provided on the screen, striving to keep the lines and values “green” in guiding their landing mechanics. Additionally, knee flexion angle was presented in accordance with the same conventions. The control group performed an identical 31 cm drop vertical jump landing protocol, at the same frequency and intensity, but without the Kinect-based real-time feedback.

Fig. 3. Subject interacting with KBF system during training intervention.

All subjects were read the same instructional statement prior to each training session which emphasized the key components of proper lower extremity landing mechanics. This statement was, “When performing your jumps, try to land softly and balanced.” Accordingly, in order to reinforce the verbal statement, a demonstration of proper technique was performed by trained research team member for observation by all participants on the first day of the study. Any propensity for detrimental over-compensatory responses to feedback with respect to excessive frontal plane foot separation (hip abduction) was mitigated by providing a “not to exceed” marked landing zone, as visually represented with tape lines on the lateral borders of the landing area. This width was set to three times the anatomical average adolescent female hip width (70 mm). All sessions were monitored by a member of the research team. Subjects unable to complete a minimum of seventy percent of the available training sessions would have been dropped from the study, though none fell into this category. Upon completion of the four week training period, all participants were re-screened utilizing the same data collection methods, again without feedback, and data were stored for comparison to pre-test data. 2.5. Data processing and statistical analysis All pre-training and post-training Kinect-acquired raw data were captured at 30 Hz and exported to Excel (Microsoft, Redmond, WA, USA) where they were smoothed utilizing a nine-point moving average window. Peak knee flexion angle and minimum knee separation distance values were identified as the single peak value occurring within the first 50 ms following initial ground contact for each trial. All peak values were averaged for each subject across trials. Peak knee separation distance was normalized by subject height in meters to enable appropriate comparisons. A factorial MANCOVA was performed for all main outcome measures. A MANCOVA approach was selected to account for any potential between-group variations in pre-test baseline kinematic values. Dependent variables for this study were peak knee flexion angle in the sagittal plane and minimum knee separation distance in the frontal plane. Independent variables were “Group” (Kinect-based feedback vs. control) and “Time” (pre-training vs. post-

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx

training). All statistical analyses were performed using SPSS version 17.0 (IBM, NY, USA). Alpha-level was set a priori at p ≤ 0.05 for all statistical comparisons. All values are reported as the mean (SD), unless otherwise indicated. 3. Results

5

group (training) change in minimum knee separation distance with the KBF group experiencing significantly greater increase in frontal plane knee separation (+ 32.4 mm m− 1 vs. + 12.1 mm m− 1, p = 0.024) as compared with the control group. The effect size of this statistic was moderate (d = 0.99, r = 0.44) with a power of 0.73. Knee separation distance increased 21% in the KBF training program as compared with an 8% increase in the control group (Fig. 4b).

3.1. Baseline demographics 4. Discussion A complete summary of group anthropomorphic data, baseline kinematics, and outcome measures is presented in Table 1. Though there was a significant difference in subject age (p = 0.03) between groups, this was considered largely inconsequential as there was no significant difference in competitive gymnastics ability level (p = 0.664), height (p = 0.229), or body mass (p = 0.211) between control and KBF groups. Additionally, there was no significant difference in number of training sessions completed (p = 0.905). Appropriately, there was no significant difference between groups in pre-training peak knee flexion angle (p = 0.146) or peak minimum knee separation distance (p = 0.855). 3.2. Peak knee flexion angle There was a significant within-group (time) improvement in sagittal plane peak knee flexion angle (pre-test: 12.5(4.3)° vs. post-test: 19.4(5.5)°, p = 0.002) for the KBF training group. The effect size of this statistic was large (d = 1.40, r = 0.570) which yielded a power of 0.90. The control training group, however, experienced no significant increase in peak knee flexion angle (pre-test: 15.2(4.4)° vs. post-test: 14.8(3.9)°, p = 0.614). There was a between-group change in peak knee flexion angle with the KBF group showing a significantly greater increase in knee flexion angle (KBF: + 6.4° vs. CTRL: − 0.4°, p = 0.001) as compared with the control group. The effect size of this statistic was large (d = 1.618, r = 0.629) which yielded a power of 0.93. Knee flexion increased 46% in the Kinect-based biofeedback training group, while virtually no change occurred in the control group (Fig. 4a).

As supported by the literature (Hewett et al., 2006b; Shultz, 2008; Quatman and Hewett, 2009), aberrant knee kinematics may be indicative of increased risk of non-contact ACL injury in the female athlete. Separately, existing literature supports the efficacy of visual feedback as a catalyst for inducing neuromuscular motor control changes in functional mechanics (Benjaminse et al., 2015). It has been suggested that gradual removal of real-time feedback over the course of the intervention may prolong the neuromuscular training benefits (Hewett et al., 2006b). Crowell and Davis, in their 2011 study of real-time feedback methods on reducing lower extremity loads, found that gradual removal of the feedback prolonged the beneficial load reducing training effects (Hewett et al., 2006b). The results of the present study demonstrate that feedback training with a Kinect-based system (KBF) significantly improved knee kinematics associated with increased risk of non-contact ACL injury. Specifically, the group that received KBF training adopted a significantly greater knee flexion angle and knee separation distance during landing as compared with the group that did not receive feedback during the DVJ. In some cases, this improvement was also visually and qualitatively apparent. Since there is evidence linking the combination of excessive knee valgus in the frontal plane and insufficient knee flexion angle in the sagittal plane to an increased predisposition for non-contact ACL injury in female athletes (Hewett et al., 2004, 2005a; Quatman and Hewett, 2009), these findings provide promise for KBF as a potential non-contact ACL injury prevention tool. 4.1. Clinical relevance

3.3. Minimum knee separation distance There was a significant within-group (time) improvement in subject height-normalized frontal plane knee separation distance for the KBF training group (pre-test: 152.8(16.8) mm m−1 vs. post-test: 185.2(23.3) mm m−1, p b 0.001) as compared with the control training group (pre-test: 154.4(24.2) mm m− 1 vs. post-test: 166.3(22.4) mm m− 1, p = 0.064). The effect size of this statistic was large (d = 1.60, r = 0.620) with a power of 0.98. There was also a between-

Table 1 Anthropomorphic and outcome measure group means (SD), and statistical differences. CTRL Anthropomorphic data Age in years (SD) Height in meters (SD) Body mass in kilograms (SD) Athletic ability level (SD) Training sessions completed (SD) (of 12 training sessions available) Outcome measures Pre peak knee flexion in degrees (SD) Post peak knee flexion in degrees (SD) Change Pre peak knee sep dist in mmm (SD) Post peak knee sep dist in mmm (SD) Change ⁎ p b 0.05. ⁎⁎ p b 0.01.

KBF

p

14.3 (1.4) 1.6 (0.1) 51.5 (6.3) 7.3 (1.3) 9.2 (1.5)

15.7 (1.6) 1.6 (0.1) 55.2 (7.6) 7.5 (0.9) 9.1 (1.8)

0.030⁎ 0.229 0.211 0.664 0.905

15.2 (4.4) 14.8 (3.9) −0.4 154.4 (24.2) 166.3 (22.4) 12.1

12.5 (4.3) 19.4 (5.5) 6.9 152.8 (16.8) 185.2 (23.3) 32.4

0.146 0.025⁎ 0.001⁎⁎ 0.855 0.056 0.024⁎

The KBF training intervention effectively altered knee kinematics consistent with non-contact ACL injury risk. The autonomous training capabilities afforded by this approach may assist clinicians and coaches by improving compliance with, and time efficiency of, injury prevention training measures within clinical and in-the-field environments. As higher levels of compliance have been correlated to decreased ACL injury rates (Sugimoto et al., 2012), intervention methods that increase compliance are desired. In this study, compliance was over seventyfive percent, which is higher than that reported in many alternative intervention approach studies (Sugimoto et al., 2012). 4.2. Limitations Although deliberate measures were employed to minimize limitations within this study, some inherent limitations were present. These included sample size, homogeneity of the sample, and kinematic signal processing errors. Sample size limitations were largely overcome for most outcome measures within this study as adequate power was reached despite reasonably small sample size. Though the sample population targeted for this study was fairly homogenous, considering that only female competitive gymnasts between the ages of 13 and 18 years were utilized, the benefits of such subject group targeting outweighed the drawbacks. Subject homogeneity afforded the capability to limit extraneous training and environmental variables that might otherwise have confounded the intervention. In addition, the use of athletes who typically train and perform without footwear eliminated a potentially confounding variable of shoe-type that would otherwise have triggered necessity of an additional battery of control measures.

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

6

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx

Change in Peak Knee Flexion (IC+50ms)

Change in Peak Knee Separation (IC + 50ms) 45

8 (degrees)

Knee Flexion Angle

SD 6.9 °

6 4 SD

2

-0.4 °

0 -2 -4 -6

KBF Group

CTRL Group

a

*p=0.024

40

SD

35

32.4 mm

30 -1)

*p=0.001 10

Knee Separation Distance

12

25 SD

20 15

12.1 mm

10 5 0

KBF Group

CTRL Group

b Fig. 4. Group comparisons for peak knee flexion (4a) and peak knee separation distance (4b).

The Kinect hardware has an upper ceiling of 30 Hz for kinematic data capture. Ideally, future efforts to access higher rate capture from such a system would be beneficial for dynamic functional movement tracking and would yield higher resolution data with less need for data-point extrapolation. With regards to kinematic signal processing errors, sources of skin artifact and physical marker inertial property errors were largely mitigated by the markerless nature of the collection system. However, the potential remained for some noise to result from pixel tracking errors within the software-based calculation of segment locations due to variations in subject morphology or body composition. However, as the subjects in this study were all of very lean body composition, this source of error was minimal and likely inconsequential.

in peak knee flexion and minimum knee separation distance during initial ground contact phase of DVJ, both of which may be associated with a reduction in the risk of non-contact ACL injury in female athletes. Considering the mean lifetime societal economic impact of an ACL injury is estimated at over $38,000 per case (Mather et al., 2013), the results of this study may guide support and guide future efforts targeting injury reduction. This study, and other such approaches (Barrios et al., 2010; Crowell and Davis, 2011; Dowling et al., 2012; Myer et al., 2013), mark a potential evolution in clinical feedbackbased training. The Kinect-based feedback intervention proved to be effective in mitigating non-contact ACL injury-predisposing knee kinematics in adolescent female athletes. Research efforts aimed at extending the applicability of this approach to the sports medicine and therapy clinical environment are warranted.

4.3. Recommendations for future research Conflict of interest Though the results of this study demonstrated that KBF training may be effective for modifying kinematics associated with an increased risk of non-contact ACL injury in adolescent female athletes, further research is warranted in a number of areas. Since the population was limited to female gymnasts, the transferability of these findings to all female athletes, including those in other high risk sports such as soccer and basketball, should be explored in future studies. Additionally, the transferability of these findings to male athletes, as it has been well established that they tend to succumb to non-contact ACL tears in a very different manner biomechanically (Hewett et al., 2006a), will be investigated. In future work, dynamic knee valgus for each leg separately should be addressed. While knee separation distance has, as aforementioned, been used as a surrogate for knee valgus on bilateral activities, some individuals may have innately developed a functional motor paradigm that enables them to perform drop landings with properly aligned low-valgus landing technique coupled with small normalized knee separation distances. Though significant improvements in biomechanics consistent with decreased risk of non-contact ACL injury were observed, the short and long-term duration of these improvements, once the real-time feedback measures are withdrawn from the training environment, must be determined. The full impact of KBF training on the rate of acute ACL tears will require a longitudinal epidemiological study of a sufficient sample size over an extended period of time. 5. Conclusion After four weeks of real-time feedback training, the subjects in this controlled environment study demonstrated significant improvements

No conflicts of interest are present for any of the authors who contributed to this study. No external sources of funding were received for this study. Acknowledgments The authors wish to thank the participating athletes, coaches, and administrators for their time and cooperation with this study. Additionally, the authors wish to thank Dr. Vijay K. Goel, Dr. Barry Scheuermann, and Dr. Martin Rice for their insight into study design and statistical analysis. No external sources of funding were received for this study. References Agel, J., Arendt, E.A., Bershadsky, B., 2005. Anterior cruciate ligament injury in national collegiate athletic association basketball and soccer: a 13-year review. Am. J. Sports Med. 33 (4), 524–530. http://dx.doi.org/10.1177/0363546504269937 (PubMed PMID: 15722283). Barrios, J.A., Crossley, K.M., Davis, I.S., 2010. Gait retraining to reduce the knee adduction moment through real-time visual feedback of dynamic knee alignment. J. Biomech. 43 (11), 2208–2213. http://dx.doi.org/10.1016/j.jbiomech.2010.03.040. Benjaminse, A., Lemmink, K.A., Diercks, R.L., Otten, B., 2010. An investigation of motor learning during side-step cutting: design of a randomised controlled trial. BMC Musculoskelet. Disord. 11, 235. http://dx.doi.org/10.1186/1471-2474-11-235 (PubMed PMID: 20942920; PMCID: 2976735). Benjaminse, A., Gokeler, A., Dowling, A.V., Faigenbaum, A., Ford, K.R., Hewett, T.E., Onate, J.A., Otten, B., Myer, G.D., 2015. Optimization of the anterior cruciate ligament injury prevention paradigm: novel feedback techniques to enhance motor learning and reduce injury risk. J. Orthop. Sports Phys. Ther. 45 (3), 170–182. http://dx.doi.org/ 10.2519/jospt.2015.4986 (PubMed PMID: 25627151). Borenstein, G., 2012. Making Things See: 3D Vision with Kinect. Processing, Arduino, and MakerBot: Make (440 p.). Brindle, T.J., Mizelle, J.C., Lebiedowska, M.K., Miller, J.L., Stanhope, S.J., 2009. Visual and proprioceptive feedback improves knee joint position sense. Knee Surg. Sports

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018

E. Nyman Jr., C.W. Armstrong / Clinical Biomechanics xxx (2015) xxx–xxx Traumatol. Arthrosc. 17 (1), 40–47. http://dx.doi.org/10.1007/s00167-008-0638-3 (PubMed PMID: 18839143; PMCID: 3629544). Chaudhari, A.M., Lindenfeld, T.N., Andriacchi, T.P., Hewett, T.E., Riccobene, J., Myer, G.D., Noyes, F.R., 2007. Knee and hip loading patterns at different phases in the menstrual cycle: implications for the gender difference in anterior cruciate ligament injury rates. Am. J. Sports Med. 35 (5), 793–800. http://dx.doi.org/10.1177/0363546506297537 (PubMed PMID: 17307891). Crowell, H.P., Davis, I.S., 2011. Gait retraining to reduce lower extremity loading in runners. Clin. Biochem. 26 (1), 78–83. http://dx.doi.org/10.1016/j.clinbiomech.2010. 09.003 (PubMed PMID: 20888675; PMCID: 3014399). Dallinga, J.M., Benjaminse, A., Leinmink, K.A.P.M., 2012. Which screening tools can predict injury to the lower extremities in team sports?: a systematic review. Sports Med. 42 (9), 791–815 (PubMed PMID: 79746792). DiStefano, L.J., Blackburn, J.T., Marshall, S.W., Guskiewicz, K.M., Garrett, W.E., Padua, D.A., 2011. Effects of an age-specific anterior cruciate ligament injury prevention program on lower extremity biomechanics in children. Am. J. Sports Med. 39 (5), 949–957. http://dx.doi.org/10.1177/0363546510392015 (PubMed PMID: 21285445). Donnelly, C.J., Lloyd, D.G., Elliott, B.C., Reinbolt, J.A., 2012. Optimizing whole-body kinematics to minimize valgus knee loading during sidestepping: implications for ACL injury risk. J. Biomech. 45 (8), 1491–1497. http://dx.doi.org/10.1016/j.jbiomech. 2012.02.010 (PubMed PMID: 22387123). Dowling, A.V., Favre, J., Andriacchi, T.P., 2012. Inertial sensor-based feedback can reduce key risk metrics for anterior cruciate ligament injury during jump landings. Am. J. Sports Med. 40 (5), 1075–1083. http://dx.doi.org/10.1177/0363546512437529. Dutta, T., 2012. Evaluation of the Kinect sensor for 3-D kinematic measurement in the workplace. Appl. Ergon. 43 (4), 645–649. http://dx.doi.org/10.1016/j.apergo.2011. 09.011 (PubMed PMID: 22018839). Ford, K.R., Myer, G.D., Toms, H.E., Hewett, T.E., 2005a. Gender differences in the kinematics of unanticipated cutting in young athletes. Med. Sci. Sports Exerc. 37 (1), 124–129 (PubMed PMID: 15632678). Ford, K.R., Myer, G.D., Smith, R.L., Byrnes, R.N., Dopirak, S.E., Hewett, T.E., 2005b. Use of an overhead goal alters vertical jump performance and biomechanics. J. Strength Cond. Res. 19 (2), 394–399. http://dx.doi.org/10.1519/00124278-200505000-00026 (PubMed PMID: WOS:000229411100026). Grindstaff, T.L., Hammill, R.R., Tuzson, A.E., Hertel, J., 2006. Neuromuscular control training programs and noncontact anterior cruciate ligament injury rates in female athletes: a numbers-needed-to-treat analysis. J. Athl. Train. 41 (4), 450–456 (PubMed PMID: 17273472; PMCID: 1748422). Herman, D.C., Onate, J.A., Weinhold, P.S., Guskiewicz, K.M., Garrett, W.E., Yu, B., Padua, D.A., 2009. The effects of feedback with and without strength training on lower extremity biomechanics. Am. J. Sports Med. 37 (7), 1301–1308. http://dx.doi.org/ 10.1177/0363546509332253 (PubMed PMID: 19299530). Hewett, T.E., 2008. Predisposition to ACL injuries in female athletes versus male athletes. Orthopedics 31 (1), 26–28 (PubMed PMID: 18269164). Hewett, T.E., Myer, G.D., Ford, K.R., 2004. Decrease in neuromuscular control about the knee with maturation in female athletes. J. Bone Joint Surg. Am. 86-A (8), 1601–1608 (PubMed PMID: 15292405). Hewett, T.E., Myer, G.D., Ford, K.R., Heidt Jr., R.S., Colosimo, A.J., McLean, S.G., van den Bogert, A.J., Paterno, M.V., Succop, P., 2005a. Biomechanical measures of neuromuscular control and valgus loading of the knee predict anterior cruciate ligament injury risk in female athletes: a prospective study. Am. J. Sports Med. 33 (4), 492–501. http://dx.doi.org/10.1177/0363546504269591 (PubMed PMID: 15722287). Hewett, T.E., Zazulak, B.T., Myer, G.D., Ford, K.R., 2005b. A review of electromyographic activation levels, timing differences, and increased anterior cruciate ligament injury incidence in female athletes. Br. J. Sports Med. 39 (6), 347–350. http://dx.doi.org/ 10.1136/bjsm.2005.018572 (PubMed PMID: 15911605; PMCID: 1725222). Hewett, T.E., Myer, G.D., Ford, K.R., 2006a. Anterior cruciate ligament injuries in female athletes: Part 1, mechanisms and risk factors. Am. J. Sports Med. 34 (2), 299–311. http://dx.doi.org/10.1177/0363546505284183 (PubMed PMID: 16423913). Hewett, T., Myer, G., Ford, K.R., JR, S., 2006b. Preparticipation physical examination using a Box drop vertical jump test in young athletes. Clin. J. Sport Med. 16 (4). Hootman, J.M., Albohm, M.J., 2012. Anterior cruciate ligament injury prevention and primary prevention of knee osteoarthritis. J. Athl. Train. 47 (5), 589–590. http://dx. doi.org/10.4085/1062-6050-47.5.06 (PubMed PMID: 23068596; PMCID: 3465039). Hughes, G., Watkins, J., 2006. A risk-factor model for anterior cruciate ligament injury. Sports Med. 36 (5), 411–428 (PubMed PMID: 16646629). Kiefer, A.W., Kushner, A.M., Groene, J., Williams, C., Riley, M.A., Myer, G.D., 2015. A commentary on real-time biofeedback to augment neuromuscular training for ACL injury prevention in adolescent athletes. J. Sports Sci. Med. 14 (1), 1–8 (PubMed PMID: 25729282; PMCID: 4306760). Koga, H., Nakamae, A., Shima, Y., Iwasa, J., Myklebust, G., Engebretsen, L., Bahr, R., Krosshaug, T., 2010. Mechanisms for noncontact anterior cruciate ligament injuries: knee joint kinematics in 10 injury situations from female team handball and basketball. Am. J. Sports Med. 38 (11), 2218–2225. http://dx.doi.org/10.1177/0363546510373570 (PubMed PMID: 20595545). Kramer, L.C., Denegar, C.R., Buckley, W.E., Hertel, J., 2007. Factors associated with anterior cruciate ligament injury: history in female athletes. J. Sports Med. Phys. Fitness 47 (4), 446–454 (PubMed PMID: 18091686).

7

Mather 3rd, R.C., Koenig, L., Kocher, M.S., Dall, T.M., Gallo, P., Scott, D.J., Bach Jr., B.R., Spindler, K.P., Group, M.K., 2013. Societal and economic impact of anterior cruciate ligament tears. J. Bone Joint Surg. Am. 95 (19), 1751–1759. http://dx.doi.org/10. 2106/JBJS.L.01705 (PubMed PMID: 24088967; PMCID: 3779900). McNair, P.J., Marshall, R.N., Matheson, J.A., 1990. Important features associated with acute anterior cruciate ligament injury. N. Z. Med. J. 103 (901), 537–539 (PubMed PMID: 2243642). Mentiplay, B.F., Clark, R.A., Mullins, A., Bryant, A.L., Bartold, S., Paterson, K., 2013. Reliability and validity of the Microsoft Kinect for evaluating static foot posture. J. Foot Ankle Res. 6 (1), 14. http://dx.doi.org/10.1186/1757-1146-6-14 (PubMed PMID: 23566934; PMCID: 3639226). Myer, G.D., Ford, K.R., Hewett, T.E., 2005. The effects of gender on quadriceps muscle activation strategies during a maneuver that mimics a high ACL injury risk position. J. Electromyogr. Kinesiol. 15 (2), 181–189. http://dx.doi.org/10.1016/j.jelekin.2004. 08.006 (PubMed PMID: 15664147). Myer, G.D., Stroube, B.W., DiCesare, C.A., Brent, J.L., Ford, K.R., Heidt, R.S., Hewett, T.E., 2013. Augmented feedback supports skill transfer and reduces high-risk injury landing mechanics: a double-blind, randomized controlled laboratory study. Am. J. Sports Med. 41 (3), 669–677. http://dx.doi.org/10.1177/0363546512472977. Noyes, F.R., Barber-Westin, S.D., Fleckenstein, C., Walsh, C., West, J., 2005. The drop-jump screening test: difference in lower limb control by gender and effect of neuromuscular training in female athletes. Am. J. Sports Med. 33 (2), 197–207 (PubMed PMID: 15701605). Nyman Jr., E., 2013. The Effects of an OpenNI/Kinect-Based Biofeedback Intervention on Kinematics at the Knee During Drop Vertical Jump Landings: Implications for Reducing Neuromuscular Predisposition to Non-Contact ACL Injury Risk in the Young Female Athlete. The University of Toledo, Toledo, OH. Nyman, E., Armstrong, C.W., Scheuermann, B.W., 2013. The Effects of an OpenNI/KinectBased Biofeedback Intervention on Kinematics at the Knee During Drop Vertical Jump Landings: Implications for Reducing Neuromuscular Predisposition to NonContact ACL Injury Risk in the Young Female Athlete. University of Toledo, University of Toledo (Dissertation). Obdralek, S., 2012. Accuracy and robustness of kinect pose estimation in the context of coaching of elderly population. 2012 Annual International Conference of the IEEE. IEEE, San Diego, CA, pp. 1188–1193. Onate, J.A., Guskiewicz, K.M., Marshall, S.W., Giuliani, C., Yu, B., Garrett, W.E., 2005. Instruction of jump-landing technique using videotape feedback: altering lower extremity motion patterns. Am. J. Sports Med. 33 (6), 831–842. http://dx.doi.org/ 10.1177/0363546504271499 (PubMed PMID: 15827359). Paszkewicz, J., Webb, T., Waters, B., Welch McCarty, C., Van Lunen, B., 2012. The effectiveness of injury-prevention programs in reducing the incidence of anterior cruciate ligament sprains in adolescent athletes. J. Sport Rehabil. 21 (4), 371–377 (PubMed PMID: 22713252). Quatman, C.E., Hewett, T.E., 2009. The anterior cruciate ligament injury controversy: is “valgus collapse” a sex-specific mechanism? Br. J. Sports Med. 43 (5), 328–335. http://dx.doi.org/10.1136/bjsm.2009.059139 (PubMed PMID: 19372087). Quatman, C.E., Kiapour, A.M., Demetropoulos, C.K., Kiapour, A., Wordeman, S.C., Levine, J.W., Goel, V.K., Hewett, T.E., 2014. Preferential loading of the ACL compared with the MCL during landing: a novel in sim approach yields the multiplanar mechanism of dynamic valgus during ACL injuries. Am. J. Sports Med. 42 (1), 177–186. http://dx. doi.org/10.1177/0363546513506558 (PubMed PMID: 24124198; PMCID: 3927458). Schmitz, A., Ye, M., Shapiro, R., Yang, R., Noehren, B., 2014. Accuracy and repeatability of joint angles measured using a single camera markerless motion capture system. J. Biomech. 47 (2), 587–591. http://dx.doi.org/10.1016/j.jbiomech.2013.11.031 (PubMed PMID: 24315287). Shultz, S.J., 2008. ACL injury in the female athlete: a multifactorial problem that remains poorly understood. J. Athl. Train. 43 (5), 455. http://dx.doi.org/10.4085/1062-605043.5.455 (PubMed PMID: 18833306; PMCID: 2547863). Stroube, B.W., Myer, G.D., Brent, J.L., Ford, K.R., Heidt Jr., R.S., Hewett, T.E., 2013. Effects of task-specific augmented feedback on deficit modification during performance of the tuck-jump exercise. J. Sport Rehabil. 22 (1), 7–18 (PubMed PMID: 23238301). Sugimoto, D., Myer, G.D., Bush, H.M., Klugman, M.F., Medina McKeon, J.M., Hewett, T.E., 2012. Compliance with neuromuscular training and anterior cruciate ligament injury risk reduction in female athletes: a meta-analysis. J. Athl. Train. 47 (6), 714–723. http://dx. doi.org/10.4085/1062-6050-47.6.10 (PubMed PMID: 23182020; PMCID: 3499896). Swenson, D.M., Collins, C.L., Best, T.M., Flannigan, D.C., Fields, S.K., Comstock, R.D., 2013. Epidemiology of knee injuries among U.S. high school athletes, 2005/2006–2010/ 2011. Med. Sci. Sports Exerc. 45 (3), 462–469. http://dx.doi.org/10.1249/MSS. 0b013e318277acca. Wong, W., How Microsoft's PrimeSense-based Kinect Really Works. www.electronicdesign. com2011 (cited 2013 6/5/2013). Zazulak, B.T., Paterno, M., Myer, G.D., Romani, W.A., Hewett, T.E., 2006. The effects of the menstrual cycle on anterior knee laxity: a systematic review. Sports Med. 36 (10), 847–862 (PubMed PMID: 17004848). Zazulak, B.T., Hewett, T.E., Reeves, N.P., Goldberg, B., Cholewicki, J., 2007. Deficits in neuromuscular control of the trunk predict knee injury risk: a prospective biomechanical–epidemiologic study. Am. J. Sports Med. 35 (7), 1123–1130. http://dx.doi. org/10.1177/0363546507301585 (PubMed PMID: 17468378).

Please cite this article as: Nyman, E., Armstrong, C.W., Real-time feedback during drop landing training improves subsequent frontal and sagittal plane knee kinematics, Clin. Biomech. (2015), http://dx.doi.org/10.1016/j.clinbiomech.2015.06.018