Single-Session Video and Electromyography Feedback in Overhead Athletes With Scapular Dyskinesis and Impingement Syndrome

Study Design

This was a laboratory-based, cross-sectional study.

Participants

According to previous studies,^1,15,23 a total sample size of 40 participants was calculated to provide 80% power with effect sizes of 0.91, 0.86, and 0.94 to detect differences in scapular posterior tilt (3.7°), UT/LT ratio (31%), and LT muscle activity (5.98% of maximal voluntary isometric contraction; MVIC), respectively. Diagnosis of overhead athletes with SIS was based on positive results on at least 2 of the following 5 impingement tests: (1) Neer impingement test, (2) Hawkins-Kennedy impingement test, (3) empty can test, (4) external resisted test, (5) pain during rotator cuff palpation.²⁴ Additionally, all participants with SIS had medial border prominence dyskinesis.²⁵ Volunteers were excluded from the study if they had any of the following conditions: (1) history of shoulder dislocation, fracture, or surgery within the past year; (2) history of direct contact injury to the neck or upper extremities within the past month; (3) glenohumeral joint instability (positive apprehension test, sulcus sign); (4) neurologic disorder; or (5) pain (visual analogue scale rating > 5) during the experimental tasks. The National Taiwan University Hospital Human Subject Research Ethics Committee approved this study. All participants gave written informed consent before data collection began.

Instrumentation

The 3Space FASTRAK (Polhemus Inc, Colchester, VT), an electromagnetic motion-analysis system with Motion Monitor software (Innovative Sports Training, Inc, Chicago, IL), was used to collect 3-dimensional scapular kinematics. According to the manufacturer, the system is accurate to 0.8 mm and 0.15°. A previous validity study²⁶ showed that the FASTRAK could be used for collection of kinematic data under 120° of arm elevation in lean young participants. Three electromagnetic sensors were used for kinematic tracking of the scapula. The first sensor was attached to the sternum, and the second was attached to the flat bony surface of the acromion with adhesive tape. The third sensor was attached to the distal humerus with hook-and-loop straps. Local coordinate systems were marked with a stylus by an experienced physical therapist as follows: sternal notch, xiphoid process, seventh cervical vertebrae, eighth thoracic vertebrae, 12th thoracic vertebrae, acromion, anterior glenohumeral joint, posterior glenohumeral joint, root of the spine of the scapula, inferior angle of the scapula, lateral epicondyle, and medial epicondyle.

Following the Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscles (SENIAM) guidelines,²⁷ we arranged the surface (s)EMG assemblies to consist of pairs of silver chloride circular (recording diameter of 10 mm) surface electrodes (The Ludlow Company LP, Chicopee, MA) with an interelectrode (center-to-center) distance of 20 mm, and a Grass AC/DC amplifier (model 15A12; Astro-Med Inc, West Warwick, RI) with a gain of 1000, a common mode rejection ratio of 86 dB at 60 Hz, and a bandwidth (−3 dB) of 10 to 1000 Hz. The sEMG data were collected at 1000 Hz per channel using a 16-bit analog-to-digital converter (model MP 150; BIOPAC Systems Inc, Goleta, CA). Surface EMG electrodes were placed on the UT, LT, and SA of the involved shoulder. Electrodes for the UT were placed midway between the acromion and the seventh spinous process of the cervical vertebrae. The LT was palpated obliquely upward and laterally along the line between the intersection of the spine of the scapula and the seventh spinous process of the thoracic vertebrae. Electrodes for the SA were placed anterior to the latissimus dorsi and posterior to the pectoralis major. The reference electrode was placed on the ipsilateral clavicle. Full bandwidth sEMG data captured by the data-acquisition software (AcqKnowledge; BIOPAC Systems Inc) were reduced using a root mean square algorithm to produce sEMG envelopes with an effective sampling rate of 50 samples. A trigger point was used at the beginning of the movements to synchronize 3-dimensional kinematic data and EMG data.

Feedback Training

Participants were randomly assigned to the video feedback or EMG biofeedback group by block randomization (4 participants per block) after baseline data collection. During the experiment, male participants removed their shirts and female participants wore halter tops. Before the main experiment, surface EMG electrodes and FASTRAK kinematic sensors were attached to the participants.

The participants learned to use constant feedback to correct their scapular-muscle activation or scapular kinematics throughout the experiment. Both groups received feedback, either from video or EMG. The scapular-correction training consisted of 2 sections, learning and training. In the learning section, the participants learned to position the scapula in neutral while in the resting position (arms by sides) and were familiarized with the feedback. Scapular neutral position was defined as both scapulae being relatively symmetric with no prominence of the scapular medial border or inferior angle.²⁸ The video feedback group was instructed to decrease the prominence of the scapular inferior angle and medial border on the screen by placing the scapula tightly against the rib cage (Figure 1). The EMG biofeedback group was instructed to focus on the UT/LT muscle balance ratio signals. The goal was to keep the activity ratio below a specified threshold. The threshold was based on the range of 1 standard deviation (SD) of the pretest ratio average data (Figure 2). The threshold for scapular control learning was based on a pilot study conducted before the present study, which identified that the 1.5 SD value was too difficult for participants, whereas the 0.5 SD value was too easy. Thus, we chose 1 SD as the threshold.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

The participants practiced in the resting position until they were able to successfully achieve the stated goals: either decreasing the prominence of the medial border of the scapula (video feedback group) or lowering the magnitude of muscle ratios below the threshold (EMG biofeedback group) 3 times in a row. When the participants were able to correct the scapula to neutral while resting, they were instructed to proceed to the training section to correct the scapula during arm elevation and lowering. Their goals were also to decrease the prominence of the scapula on the video feedback or to maintain the UT/LT muscle balance ratio signals below a threshold. The arm-elevation and -lowering task included arm elevation in the scapular plane to 180° within 3 seconds and lowering within 3 seconds. During arm elevation and lowering, male participants held a 1-kg dumbbell and female participants held a 0.5-kg dumbbell. Oral cues such as “Retract the scapula,” “Do not overshrug the shoulder,” “Relax your shoulders,” and “Slightly squeeze the muscles between the scapula” and tactile cues were given to the participants as supplemental instructions in both groups throughout the experiment. Participants were allowed to practice for 9 trials in 3 blocks of 3 trials per block. A 1-minute rest period was allowed between blocks.

Time-Point Measures

Muscle activity (UT, LT, SA), muscle balance ratios (UT/LT, UT/SA), and scapular kinematics (anterior-posterior tilt, external-internal rotation, upward rotation) were collected during the arm-elevation and -lowering task. The arm-elevation and -lowering task was conducted 3 times each before and after the feedback intervention with and without the feedback condition.

Data Reduction

The raw kinematic data were low-pass filtered at a 6-Hz cutoff frequency and converted into anatomically defined rotations. In general, we followed the International Society of Biomechanics guidelines (https://isbweb.org/images/documents/standards/Wu%20et%20al%20J%20Biomech%2038%20(2005)%20981%E2%80%93992.pdf) for constructing a shoulder-joint coordinate system. Scapular orientation relative to the thorax was described using a Euler angle sequence of rotation about Z_s (internal-external rotation), rotation about Y'_s (downward-upward rotation), and rotation about X"_s (posterior-anterior tilt). We used scapular internal rotation, upward rotation, and anterior tilt in the final analyses.

Full bandwidth sEMG data captured by data-acquisition software (AcqKnowledge) were reduced using a root mean square algorithm to produce sEMG envelopes with an effective sampling rate of 50 samples. The EMG data for each muscle were the averages of 3 trials. The mean sEMG amplitude of each muscle, reported as a percentage of MVIC, was used to display the normalized activity of the UT, LT, and SA. The MVIC for the UT muscle was measured during resisted shoulder flexion. Participants were seated with the shoulder flexed to 90° and resistance applied to the distal arm.²⁹ For the MVIC measurement of the LT muscle, participants lay prone with the test arm abducted in line with the muscle fibers. Resistance was applied against further elevation. For the MVIC measurement of the SA muscle, participants were seated with the arm elevated to 135°. Resistance was applied to the distal upper arm against further elevation.³⁰ The MVICs were collected during 3 trials of 5 seconds each, with 1-minute rest intervals between trials. Muscle balance ratios were calculated by dividing the adjusted UT muscle activity by the adjusted LT and SA muscle activity. Muscle activity was recorded during 0° to 30°, 30° to 60°, 60° to 90°, 90° to 120°, and >120° of arm elevation and lowering.

Statistical Analysis

We used SPSS (version 20.0; IBM Corp, Armonk, NY) for data analysis. The Shapiro-Wilk test was performed to confirm normal distribution of the kinematic and EMG data. Three-way mixed analysis of variance was conducted for scapular kinematics and EMG outcomes, with factors of condition (pretraining, posttraining, post without feedback), group (video feedback, EMG biofeedback), and arm-elevation phase (phases of the kinematic data at raising 30°, raising 60°, raising 90°, raising 120°, lowering 120°, lowering 90°, lowering 60°, and lowering 30°; and muscle EMG data at raising 0°–30°, raising 30°–60°, raising 60°–90°, raising 90°–120°, raising >120°, lowering >120°, lowering 120°–90°, lowering 90°–60°, lowering 60°–30°, and lowering 30°–0°). A Bonferroni correction was used to adjust the α level, which was set at .05. If the data did not meet the normality assumptions, nonparametric tests were conducted for the measures.

For nonparametric data, we used the Friedman test and Mann-Whitney U test to compare outcomes between conditions and groups. For the between-groups analyses, we compared the between-groups values for different elevation angles and time (pretraining, posttraining, post without feedback) separately; thus, the adjusted P value was .025 (.05/2). For the within-group analyses, we compared time (pretraining, posttraining, post without feedback) at different elevation angles; thus, the adjusted P value was .016 (.05/3). The r family effect sizes were used in the analyses. The scapular kinematic and EMG outcome analyses were conducted in the same manner.

Muscle Activity

The Mann-Whitney U test was conducted for between-groups differences due to nonnormal distribution. Differences were found in muscle activity and muscle balance ratios. The LT muscle activity was higher in the EMG biofeedback group during the posttraining condition in the 0° to 30°, 30° to 60° (medians = 8.84 and 21.94, P values = .001 and .002, and effect sizes = 0.30 and 0.31, respectively), and 60° to 90° (median = 29.37, P = .002, effect size = 0.32) elevation phase and the 120° to 90° lowering phase (median = 23.45, P = .01, effect size = 0.26; Table 2). In the posttraining-without-feedback condition, LT muscle activity increased in the 0° to 30° and 60° to 90° (medians = 8.67 and 26.75, P values = .011 and .011, and effect sizes = 0.27 and 0.29, respectively) raising phases in the EMG biofeedback group. In addition, in the EMG biofeedback group, the UT/LT muscle balance ratio decreased in the 30° to 60°, 60° to 90° (medians = 1.65 and 1.29, P values = .002 and .013, and effect sizes = 0.15 and 0.16, respectively), and >120° (median = 1.71, P = .013, effect size = 0.15) raising phases in the posttraining condition (Table 3).

Table 2 Lower Trapezius Muscle Activity as a Percentage of Maximal Voluntary Isometric Contraction: Median (Between-Groups Effect Sizes, Between-Conditions Effect Sizes Compared With Pretraining) [Minimum, Maximum] in Both Groups

View Fullscreen

Table 3 Upper Trapezius:Lower Trapezius Muscle Balance Ratio Medians (Between-Groups Effect Sizes, Between-Conditions Effect Sizes Compared With Pretraining) [Minimum, Maximum] in Both Groups

View Fullscreen

For the video feedback group, nonparametric analyses were used for the outcomes due to nonnormal distribution. In the posttraining conditions, UT activity (median = 35.78, P < .016, effect size = 0.10), the UT/LT muscle balance ratio (medians = 0.70–1.39, P values < .0016, effect sizes = 0.14–0.16), and UT/SA ratio (median = 0.14, P < .016, effect size = 0.13) decreased as compared with the pretraining condition. We also found declines in the post-without-feedback condition as compared with the pretraining condition, including the UT (medians = 32.24–44.22, P values < .016, effect sizes = 0.18–0.30), UT/LT (medians = 0.73–1.73, P values < .016, effect sizes = 0.12–0.21), and UT/SA (medians = 0.69–1.44, P values < .016, effect size = 0.14–0.17). The LT muscle activity increased (medians = 6.32–15.21, P values < .016, effect sizes = 0.25–0.38) during the lowering phase in the posttraining condition. In the posttraining-without-feedback condition, LT muscle activity (medians = 7.13–21.44, P values < .016, effect sizes = 0.33–0.40) increased significantly. In comparisons of posttest with feedback and without feedback, statistical significance was present for the UT/LT between the conditions (medians = 0.89–1.17, P values < .016, effect sizes = 0.11–0.17). Muscle balance ratios were lower in the posttraining-without-feedback condition than in the posttraining-with-feedback condition.

For the EMG biofeedback group, UT muscle activity (median = 29.24–39.17 and 16.15–44.22, P values < .016 and .016, and effect sizes = 0.15–0.23 and 0.10–0.21, respectively), UT/LT (medians = 0.83–1.65 and 1.08–2.29, P values < .016 and .016, and effect sizes = 0.20–0.37 and 0.15–0.31, respectively), and UT/SA (medians = 0.56–1.42 and 0.55–1.27, P values < .016 and .016, and effect sizes = 0.09–0.30 and 0.14–0.21, respectively) decreased in the posttraining and posttraining-without-feedback conditions as compared with the pretraining outcomes. The LT muscle activity was higher in the posttraining condition and posttraining-without-feedback condition (medians = 8.67–23.45 and 9.35–22.91, P values < .016 and .016, and effect sizes = 0.27–0.37 and 0.26–0.37, respectively) than in the pretraining condition. The UT/SA ratio increased during the >120° raising phase in the post-without-feedback condition (median = 0.55, P = .014, effect size = 0.10).

Scapular Kinematics

In comparisons of the 2 groups, scapular upward rotation was higher in the video group after feedback training (median = 24.29°, P = .024, effect size = 0.19; Table 4). No group differences were present in scapular anterior tilt or internal rotation.

Table 4 Scapular Upward Rotation Degree Medians (Between-Groups Effect Sizes, Between-Conditions Effect Sizes Compared With Pretraining) [Minimum, Maximum] in Both Groups

View Fullscreen

In the video feedback group, scapular internal rotation in the 30° lowering phase was lower in the posttraining condition than in the pretraining condition (median = −5.8°, P = .003, effect size = 0.20). A difference was also evident in the posttraining and posttraining-without-feedback conditions; scapular internal rotation was lower in the posttraining condition (median = −3.8°, P = .009, effect size = 0.10). However, within-group changes in scapular upward rotation and anterior tilt did not reach significant levels.

In the EMG biofeedback group, scapular upward rotation was lower in the posttraining (medians = 0.13°–30.04°, P values < .016, effect sizes = 0.19–0.22) and posttraining-without-feedback (median = 29.61°, P < .016, effect size = 0.19) conditions than in the pretraining condition. In the post-feedback condition, the degree of scapular internal rotation was decreased during the 30° lowering phase (median = −6.36°, P = .003, effect size = 0.28). No differences were demonstrated in scapular anterior tilt.