Shoulder Muscle Activity Magnitude and Timing in Individuals with Specific Scapular Movement Patterns

Abstract

Relationships between specific muscle activity alterations and specific scapular motion patterns have not been well established. The purpose of this study was to identify muscle activation magnitude and timing in two specific scapular movement groups (excessive scapular anterior tilt, scapular lateralization). Surface electromyography (EMG) was captured simultaneously with biplane video radiography and optical motion capture during two repetitions each of shoulder flexion, abduction, and unrestricted overhead reaching in 45 adults with shoulder pain. The following muscle activation magnitudes measured with EMG were assessed individually and in ratios: lower trapezius (LT), serratus anterior (SA), upper trapezius (UT), anterior deltoid (AD), and middle deltoid (MD). The timing of each muscle’s initiation was calculated as a percent of humerothoracic range of motion. For EMG magnitude, significant interactions including group were found for AD, LT/SA, and LT/AD. Findings suggest a relative reduction of SA in both groups in the constrained motions compared to the self-selected motion and a reduction in LT in the lateralization group in early motion. For EMG timing, no group effects were significant, however findings suggest delays in SA and LT activation in both groups. The findings provide evidence that individuals with different scapular movement patterns may have distinct muscle activity patterns which can inform patient-specific rehabilitation interventions.

Introduction

Conservative treatment planning for shoulder pain and movement dysfunction should be directed, at least in part, by an understanding of muscle activation driving scapular motion. Muscle activation controls muscle contraction and, therefore, controls motion. While there is evidence that biomechanical alterations in muscle activity are related to altered scapular motion patterns (Graichen et al., 2005, Lopes et al., 2015, Ludewig and Cook, 2000, Staker et al., 2022), relationships between specific muscle activity alteration and specific scapular motion patterns have not been well established.

In cases of majorly impaired neural signals to muscles—through nerve palsies and denervation—shoulder muscle activation is specifically altered and results in corresponding specific patterns of altered scapular kinematics (Crowe and Elhassan, 2016, Didesch and Tang, 2019, Kelley et al., 2008, Nawa, 2015, Roren et al., 2013, Sergides et al., 2010, Seror et al., 2018). However, these conditions are rare and do not account for the majority of shoulder dysfunction (Carenfelt and Eliasson, 1980, Eisen and Bertrand, 1972, Fardin et al., 1978). Muscle activity in people with specific scapular movement patterns in the absence of nerve palsy or denervation is critical to understand. This information can guide patient-specific rehabilitation interventions by identifying specific muscles to target based on scapular movement patterns.

Reduced scapular upward rotation is a scapular motion alteration of particular concern as upward rotation has been found to be reduced in individuals with shoulder pain compared to individuals without pain (Borstad and Ludewig, 2002, Choi and Chung, 2023, Gonçalves et al., 2021, Lawrence et al., 2014, Lin et al., 2006, Ludewig and Cook, 2000, Struyf et al., 2014). In a prospective study (Struyf et al., 2014), individuals with decreased scapular upward rotation were more likely to develop shoulder pain in a two-year follow-up period. Two sub-patterns may exist related to reduced scapular upward rotation. As upward rotation is produced by the serratus anterior and lower trapezius, these two sub-patterns represent deficient recruitment of each of those muscles. Reduced recruitment of the serratus anterior is expected to result in reduced upward rotation along with excessive scapular anterior tilt (scapular anterior tilt pattern) (Didesch and Tang, 2019, Nawa, 2015, Roren et al., 2013). Reduced recruitment of the lower trapezius is expected to result in reduced upward rotation and lateral translation of the scapula along the thorax (scapular lateralization pattern) (Crowe and Elhassan, 2016, Didesch and Tang, 2019, Kelley et al., 2008, Sergides et al., 2010).

While muscle recruitment cannot be directly measured with electromyography (EMG), muscle excitation from EMG can characterize muscular contributions both in terms of the magnitude of muscle activity during motion and the timing of muscle activity initiation relative to motion (Vigotsky et al., 2017). In addition to assessing the magnitude of the individual muscles, evaluating ratios of activity from two muscles provides a more holistic representation of underlying motor control patterns (Cools et al., 2007, de Morais Faria et al., 2008, Michener et al., 2016). In evaluating sub-patterns related to reduced upward rotation, the lower trapezius/serratus anterior ratio is of particular interest as both muscles produce scapular upward rotation, and the relative contribution of each muscle may differentiate participants into the two sub-patterns.

The purpose of this study was to identify muscle activation magnitude and timing in two specific scapular movement groups related to reduced scapular upward rotation (excessive scapular anterior tilt, scapular lateralization). Specifically, we hypothesized that the anterior tilt group would have a higher average lower trapezius/serratus anterior muscle activity ratio than the lateralization group during arm elevation. Additionally, we hypothesized that the anterior tilt group would have delayed serratus anterior muscle activation, and the lateralization group would have delayed lower trapezius muscle activation.

Methods

Sample Size Calculation

Sample size for this study was calculated a priori using a power of 80%, as is typically used in biomechanical studies, and was powered for the main effect of group. The level of significance was set to 0.05 to minimize type I error, or the likelihood of finding false positives. A meaningful group difference in muscle activity (lower trapezius/serratus anterior ratio) was determined to be 0.25 or 25% based on significant differences found between symptomatic and asymptomatic people in a previous study (Michener et al., 2016). In that study, the average standard deviation of the lower trapezius/serratus anterior ratio during arm elevation in the symptomatic group was 0.38 or 38%. Because of the specific movement pattern groupings in the current study (as compared to a broad “subacromial pain syndrome” diagnosis in the previous study), the current study’s groups were expected to be more homogenous than the generalized symptomatic and asymptomatic groups in the other study (Michener et al., 2016). Therefore, the standard deviation for this sample size calculation was set to 75% of their standard deviation, or 29%. This power analysis determined that 23 participants were needed in each group in order to detect meaningful differences between the scapular anterior tilt group and the scapular lateralization group with 80% power.

Inclusion/Exclusion Criteria

Participants had to be 18–60 years old with non-traumatic, movement-related shoulder pain lasting at least two weeks localized to the glenohumeral joint area. Pain had to be provoked by shoulder motion, and participants needed full overhead reaching range of motion. Participants with a history of shoulder trauma, surgery, known full-thickness rotator cuff tearing, or symptoms reproduced with cervical screening tests performed by a licensed physical therapist (GS) were excluded from this study. Participants with contraindications to radiation exposure or magnetic resonance imaging (MRI) were also excluded from the study. Finally, all participants needed to demonstrate one of two movement patterns related to reduced scapular upward rotation (≤35° at 120° of humerothoracic elevation) (Lawrence et al., 2014, Lawrence et al., 2019, Ludewig and Cook, 2000) during overhead arm elevation. First, the scapular anterior tilt pattern was defined as reduced upward rotation and excessive scapular anterior tilt (≥10° when their shoulder was at 90° of arm elevation in any plane of elevation) (Lawrence et al., 2014, Roren et al., 2013). Second, the scapular lateralization pattern was defined as reduced upward rotation and lateral displacement of the scapula on the thorax by ≥1 cm from rest to 90° of arm elevation in any plane (Nijs et al., 2005). Movement patterns were first assessed primarily using clinical exam techniques. Those meeting movement pattern definitions from the clinical exam proceeded to final movement pattern assessment using advanced imaging, as described below. Informed consent was obtained prior to data collection. The study was approved by the University of Minnesota Institutional Review Board.

Preliminary Movement Screening

Movement patterns for the participants’ symptomatic shoulders were initially measured using a digital inclinometer (upward rotation and anterior tilt) and digital calipers (lateral translation). Participants’ scapular measurements were taken at static positions of standing shoulder flexion and abduction (rest, 90° elevation, and 120° elevation). Fifteen participants were excluded at this stage. Participants demonstrating either the scapular anterior tilt pattern or the scapular lateralization pattern (as defined above) were offered the opportunity to proceed with the full data collection, including advanced motion analysis and EMG. Data collection for this study was done unilaterally, so for participants for whom both shoulders were symptomatic and both fit the movement pattern criteria, the more symptomatic shoulder was tested. Participants were recruited and screened until 25 participants with each movement pattern proceeded to full data collection. This oversampling was intended to allow for potential exclusion of participants after final group assignment without affecting study power.

Final Group Assignment

Participants who passed the preliminary movement screening underwent advanced kinematic analysis using biplane video radiography and 2D/3D shape-matching (humeral and scapular kinematics) and optical motion capture (thoracic kinematics) as part of a larger study described elsewhere (Saini, 2024). Biplane video radiography accuracy in our lab has been demonstrated to be 1.2–2.6° (Akbari-Shandiz et al., 2019). These data were used to precisely validate preliminary group assignments from clinical examination and ensure that groups were kinematically distinct (Saini, 2024). Five participants were excluded based on this validation, bringing study participant totals to 23 in the scapular anterior tilt group and 22 in the scapular lateralization group (total study n = 45).

EMG Instrumentation

Surface EMG was captured using the Noraxon Ultium (Noraxon USA, Scottsdale, AZ) electromyography system. This system included nine wireless EMG sensors. Each sensor was attached, via wires 3–11 inches long, to silver/silver chloride bipolar electrodes with two centimeter spacing between the two active sites. The EMG preamplifier had a common mode rejection ratio of >100 dB, reflecting the extent to which signals common to both active sites were attenuated. The preamplifier also had an input impedance of >1,000 MΩ, which was sufficiently larger than the expected impedance at the skin so that the signal was not excessively attenuated or distorted. The Ultium system had baseline noise levels of <1 μV.

Motion Analysis Instrumentation

Scapular and humeral kinematics were captured dynamically using a custom biplane video radiography system (Imaging Systems & Services, St. Painesville, OH). Thoracic kinematics were captured using an eight-camera optical motion capture system (Vero 1.3 (5 cameras) and 2.3 (3 cameras), Vicon Motion Systems, Centennial, CO). The Vicon system integrated with Vicon Nexus and MotionMonitor (Innovative Sports Training, Chicago, IL) software which allowed for synchronization of optical motion capture, radiographic, and EMG data collection.

Preparation

The participant’s skin was prepared for data collection by wiping with alcohol wipes. Adhesive stickers were used to place surface electrodes as described in Table 1 on the following muscles: lower trapezius, serratus anterior, upper trapezius, anterior deltoid, and middle deltoid.

Table 1:

Muscle-specific electromyography electrode placements and maximum voluntary contraction positions.

Muscle	Electrode Placement	MVC Position
Lower Trapezius	⅔ of the way on a line from the root of the spine of the scapula to the T8 spinous process in the direction of the line between T8 and the acromion (Hermens et al., 1999)	Participant prone with the arm overhead in line with lower trapezius muscle fibers; resistance applied at the distal forearm (Ekstrom et al., 2005, Kendall and Kendall, 1949, Maenhout et al., 2010)
Serratus Anterior	Along the mid-axillary line at ribs 6–8 (Ekstrom et al., 2005) such that the electrodes were anterior to the latissimus dorsi muscle when the arm was flexed to 90° (Hermens et al., 1999, Ludewig and Cook, 2000); oriented parallel to the direction of muscle fibers (running from the ribs to the anterior aspect of the medial scapular border)	Participant seated with no back support, shoulder flexed to 125° and elbow flexed to 90°; resistance applied to the elbow attempting to retract and downwardly rotate the scapula (Ekstrom et al., 2005, Kendall and Kendall, 1949)
Upper Trapezius	Halfway between the acromion and the C7 spinous process in the direction of the line between the acromion and C7 (Hermens et al., 1999)	Participant seated with no back support, shoulder abducted to 90°; neck side-bent toward the tested shoulder, rotated toward the opposite shoulder, then extended as resistance was applied at the head and distal humerus (Ekstrom et al., 2005, McLean et al., 2003)
Anterior Deltoid	One finger-width distal and anterior to the acromion in the direction of the line from the acromion to the thumb (Hermens et al., 1999)	Shoulder flexed to 90° with neutral rotation, elbow flexed to 90°; resistance applied at distal humerus in inferior direction (Franke et al., 2015, Kendall and Kendall, 1949)
Middle Deltoid	On the line from the acromion to the lateral epicondyle on the greatest deltoid muscle bulge in the direction of the line from the acromion to the hand (Hermens et al., 1999)	Shoulder abducted to 90° with neutral rotation, elbow flexed to 90°; resistance applied at distal humerus in inferior direction (Gurney et al., 2016, Kendall and Kendall, 1949, Reinold et al., 2004)

EMG and Motion Collection

EMG was captured simultaneously with radiographic and optical motion capture data for two repetitions each for the following seated shoulder elevation motions: shoulder flexion, shoulder abduction, and unrestricted reaching in which the participant was instructed to simulate reaching for an object on a high shelf but was otherwise free to move in a self-selected manner. EMG data were collected at 2160 Hz while Vicon optical motion capture cameras collected at 120 Hz and video radiography was captured at 60 Hz. The simultaneous collection allowed for extraction of EMG data at known shoulder positions, facilitating comparisons within and across participants. Participants were instructed to move at a consistent pace of two seconds raising the arm and two seconds lowering the arm, which minimized confounding effects of velocity on EMG comparisons (Vigotsky et al., 2017).

Prior to the arm elevation trials, EMG recordings during maximal voluntary contractions (MVCs) were captured in the muscle-specific positions described in Table 1, manually resisted by the primary investigator (GS). Each MVC was performed for two repetitions of three seconds. Participants cycled through the set of positions twice with rest breaks of 30 seconds between each collection in order to minimize the effect of fatigue. In addition to those listed in Table 1, MVC recordings were also obtained during the following (determined after pilot testing):

1. Static scapular plane abduction: shoulder placed in 90° of scapular plane abduction with neutral rotation, elbow extended; resistance applied at distal forearm in inferior direction

2. Dynamic scapular plane abduction: starting with the arm relaxed at their side, the participant elevated their arm with neutral rotation in the scapular plane until end range elevation; resistance applied at distal forearm in inferior direction throughout motion

Filtering

EMG data were filtered within Noraxon MR3 software (Noraxon USA, Scottsdale, AZ). Raw EMG data for all muscles were band pass filtered (10–500 Hz) to remove non-physiological signal and noise. Left-sided serratus anterior data were also filtered with a bidirectional high pass Butterworth infinite impulse response filter with a cutoff of 55 Hz, determined through pilot testing, to reduce cardiac signal noise. Finally, data were root mean square averaged across 30 ms time windows (Farfán et al., 2010).

Kinematic Data Processing

Scapulothoracic and humerothoracic Euler/Cardan angles were extracted using rotation sequences following the International Society of Biomechanics standards (Wu et al., 2005): scapulothoracic YX’Z” (internal/external rotation, upward/downward rotation, anterior/posterior tilt) and humerothoracic YX’Y” (plane of elevation e.g., flexion or abduction, elevation, internal/external rotation). Glenohumeral angles were extracted using an XZ’Y” sequence (elevation, plane of elevation, internal/external rotation) (Phadke et al., 2011).

EMG Normalization and Ratios

EMG data for all participants were normalized to each participant’s MVCs, allowing for comparison across individuals. Pilot testing revealed that a given muscle’s maximum activity often occurred in a trial that did not correspond to its manual muscle testing position (e.g., maximum anterior deltoid activity occurring in the serratus anterior manual muscle test position). Because of this, for each muscle, the peak magnitude muscle activity across any 250 ms window from any MVC trial (after filtering) was used for analysis. Filtered EMG data for each muscle were normalized by dividing by the respective muscle’s MVC, and data were then expressed as a percentage of maximum. Normalized EMG data were averaged within each of the following humerothoracic elevation range of motion intervals: 31–60°, 61– 90°, 91–120°, and 121–150°. These intervals were selected to ensure that all participants had data within the full analyzed range. Due to the shape of the thorax, for most individuals with their arm at their side their elevation angle starts at 10–20° and full elevation is approximately 150° (Lawrence et. al, 2014, Kolz et. al, 2022). Data from ranges of motion outside of these intervals (e.g., ≤30° or ≥151°) were obtained when present but not included in statistical analyses. The interval data were averaged for both trials of each shoulder elevation motion (flexion, abduction, and unrestricted overhead reaching). Finally, ratios of two muscles’ normalized activity were calculated by dividing one normalized muscle activity by another for each interval and each motion. In total, the following muscles and ratios were assessed in this study: lower trapezius, serratus anterior, upper trapezius, anterior deltoid, middle deltoid, lower trapezius/serratus anterior, lower trapezius/upper trapezius, serratus anterior/upper trapezius, lower trapezius/anterior deltoid, serratus anterior/anterior deltoid, lower trapezius/middle deltoid, and serratus anterior/middle deltoid.

EMG Timing

The timing of each muscle’s initiation was described as a percent of the range of motion where 0% defined the arm at the side and 100% defined end range humerothoracic elevation. For example, if a participant completed a trial of shoulder abduction with a resting humerothoracic elevation of 10° and an ending elevation of 160°, their motion would be scaled such that 10° equaled 0% and 160° equaled 100%. The encompassing 150° arc of motion would be equally distributed from 0–100%, such that each percent would represent 1.5°. For each trial, initiation of muscle activity was defined as the percent of range of motion at which muscle activity of a given muscle exceeded 5% of its MVC. This cutoff was determined after pilot testing using methods previously described by Wickham et al. (2010) in which onset was visually determined as the first detectable rise from the fluctuating baseline signal.

Pain and Function Assessment

All participants provided subjective shoulder function information using the Disabilities of the Arm, Shoulder, and Hand (DASH) outcome measure. Participants were also asked to rate their highest and lowest levels of pain over the month prior to data collection on a scale of 0 (no pain) to 100 (maximum pain imaginable).

Statistical Analysis

All statistical analyses were run using R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing; Vienna, Austria) using an alpha level of 0.05. Data were first tested for normality and homoskedasticity through visual assessment of residual plots, Shapiro-Wilk tests, and Levene’s tests. If data were not normally distributed and/or were heteroskedastic, data transformation was attempted using log transformations, Box-Cox transformations, or square root transformations. If no transformations were successful, non-parametric Wilcoxon rank sum tests were performed for each motion to compare groups. In such cases, Bonferroni p value adjustments were used to adjust for repeated tests resulting in an adjusted alpha level of 0.02.

Demographics Comparison

The two groups were compared in age, body mass index, DASH score, and pain level using t tests. Sex and dominance of the side tested were compared using chi-square tests.

EMG Magnitude

For normal data, three-way 2×4×3 (group, range of motion interval, motion respectively) mixed-model ANOVAs with a between-subjects factor of group (scapular anterior tilt group, scapular lateralization group) and within-subjects factors of range of motion interval (31–60°, 61–90°, 91–120°, 121–150°) and motion (flexion, abduction, unrestricted overhead reaching) were run. An ANOVA was run for each muscle and ratio of muscles. When interpreting the ANOVAs, three-factor interactions were assessed first. If the three-factor interaction was significant, separate two-way mixed-model ANOVAs (factors of group and interval) were performed for each motion. If there were significant two-way interactions, multiple comparison tests were performed pairwise for the means of each appropriate combination of group + interval or group + motion. If no interactions were significant, main effects were analyzed. Any significant main effects of interval or motion were followed up with multiple comparisons tests, using the Tukey HSD method, to determine which means were different from each other. The Tukey HSD method was selected because it is designed for post hoc comparisons and provides a balance of protecting against type I error while still increasing power to detect differences (Portney and Watkins, 2009). For all ANOVAs, the effects of interest were the main effect of group or the interaction of group with range of motion interval or motion.

EMG Timing

A three-way 2×5×3 (group, muscle, motion respectively) mixed-model ANOVA with a between-subjects factor of group and within-subjects factors of muscle and motion was run for muscle activity initiation. The three-factor interaction was assessed first. If the three-factor interaction was significant, separate two-way mixed-model ANOVAs (factors of group and muscle) were performed for each motion. If there were significant two-way interactions, multiple comparison tests were performed pairwise for the means of each appropriate combination of group + muscle or group + motion. If no interactions were significant, main effects were analyzed.

Results

The two groups were not statistically significantly different from each other in age, body mass index, sex, dominance of the side tested, DASH score, or pain ratings (Table 2).

Table 2:

Comparison of demographics between groups. Data are presented as mean ± standard deviation or counts. p values are derived from t tests for numeric variables and chi-square test for categorical variables. DASH = the Disabilities of the Arm, Shoulder, and Hand outcome measure. DASH and pain scores range from 0 (no disability/pain) to 100 (maximum disability/pain).

	Anterior Tilt	Lateralization	p value
Age (years)	28.91 ± 11.52	29.41 ± 10.41	0.88
Body Mass Index (kg/m2)	26.63 ± 3.55	26.28 ± 5.69	0.81
Sex	Female: 12 Male: 11	Female: 11 Male: 11	0.88
Dominance of Side Tested	Dominant: 18 Non-dominant: 5	Dominant: 15 Non-dominant: 7	0.44
DASH (0–100)	11.95 ± 6.43	12.89 ± 9.14	0.73
Lowest Pain in Past Month (0–100)	9.0 ± 12.9	7.6 ± 7.7	0.66
Highest Pain in Past Month (0–100)	47.9 ± 22.6	46.0 ± 16.2	0.75

EMG Magnitude

Normalized muscle activity magnitude data were not normally distributed and were heteroskedastic. Serratus anterior, lower trapezius/serratus anterior, serratus anterior/anterior deltoid, and lower trapezius/middle deltoid data were successfully transformed using log transformations. Lower trapezius, upper trapezius, anterior deltoid, lower trapezius/upper trapezius, serratus anterior/upper trapezius, lower trapezius/anterior deltoid, and serratus anterior/middle deltoid were successfully transformed using Box-Cox transformations. Middle deltoid data transformation was not successful, so Wilcoxon rank sum tests were performed for each motion to compare groups. Middle deltoid group comparisons were not significant for any motion (abduction: W = 4380, p = 0.34; flexion: W = 4489, p = 0.21; unrestricted reaching: W = 4488, p = 0.16). Remaining ANOVA results of EMG magnitude data are presented in Table 3. EMG magnitude means by muscle, motion, and range of motion interval are provided in Supplementary Tables A and B.

Table 3:

Analysis of variance (ANOVA) results for muscle activity magnitude. Significant results (p < 0.05) are in bold and denoted with *. Middle deltoid data were not normal and were heteroskedastic. Data transformation was unsuccessful, so an ANOVA was not run. ROM = range of motion.

Muscle Activity	Group × Motion × ROM Interval	Group × Motion	Group × ROM Interval	Group Main Effect
Lower Trapezius	F6,472 = 0.71, p = 0.65	F2,472 = 2.72, p = 0.07	F3,472 = 0.76, p = 0.52	F1,43 = 0.51, p = 0.48
Serratus Anterior	F6,472 = 0.67, p = 0.67	F2,472 = 1.94, p = 0.15	F3,472 = 0.42, p = 0.74	F1,43 = 0.03, p = 0.87
Upper Trapezius	F6,461 = 0.16, p = 0.99	F2,461 = 1.56, p = 0.21	F3,461 = 0.84, p = 0.47	F1,42 < 0.01, p = 0.97
Anterior Deltoid	F6,472 = 0.11, p = 1.00	F2,472 = 1.27, p = 0.28	F3,472 = 2.83, p = 0.04*	N/A
Middle Deltoid	N/A	N/A	N/A	N/A
Lower Trapezius/Serratus Anterior	F6,472 = 0.58, p = 0.74	F2,472 = 4.30, p = 0.01*	F3,472 = 1.42, p = 0.24	N/A
Lower Trapezius/Upper Trapezius	F6,461 = 0.37, p = 0.90	F2,461 = 2.97, p = 0.05	F3,461 = 0.14, p = 0.94	F1,42 = 0.13, p = 0.72
Serratus Anterior/Upper Trapezius	F6,461 = 0.40, p = 0.88	F2,461 = 2.66, p = 0.07	F3,461 = 0.77, p = 0.51	F1,42 < 0.01, p = 0.98
Lower Trapezius/Anterior Deltoid	F6,472 = 0.52, p = 0.79	F2,472 = 2.94, p = 0.05	F3,472 = 3.01, p = 0.03*	N/A
Serratus Anterior/Anterior Deltoid	F6,472 = 0.40, p = 0.88	F2,472 = 0.44, p = 0.64	F3,472 = 0.50, p = 0.69	F1,43 = 0.64, p = 0.43
Lower Trapezius/Middle Deltoid	F6,472 = 0.70, p = 0.65	F2,472 = 2.31, p = 0.10	F3,472 = 1.29, p = 0.28	F1,43 = 0.21, p = 0.65
Serratus Anterior/Middle Deltoid	F6,472 = 0.21, p = 0.97	F2,472 = 0.65, p = 0.53	F3,472 = 0.27, p = 0.85	F1,43 = 0.81, p = 0.37

There was a significant interaction of group and range of motion interval for anterior deltoid. The anterior tilt group progressively increased in anterior deltoid activity across range of motion intervals to a greater extent than the lateralization group (Figure 1). Pairwise statistically adjusted comparisons of group and range of motion interval did not reveal any significant meaningful comparisons (e.g., anterior deltoid activity from 31–60° in the anterior tilt group vs. the lateralization group). Back-transformed data are plotted in Supplementary Figure A.

Figure 1:

Normalized anterior deltoid muscle activity by range of motion interval and group (across motions). The interaction of group and interval was statistically significant (p = 0.04), but pairwise comparisons of group at each interval were not. Data are presented as percentage of maximum voluntary contraction (MVC) and were Box-Cox transformed. Error bars represent standard error.

A significant interaction of group and motion was found for the ratio of lower trapezius/serratus anterior. The anterior tilt group had relatively increased lower trapezius activity or relatively decreased serratus anterior activity (i.e., a more positive ratio log transformed) compared to the lateralization group in abduction and flexion (Figure 2). In unrestricted reaching, both groups had less lower trapezius activity compared to serratus anterior (log ratio <0), and this occurred to a greater extent in the anterior tilt group. Pairwise comparisons using the Tukey HSD method did not reveal any significant comparisons of group within any motion (e.g., lower trapezius/serratus anterior activity in flexion in the anterior tilt group vs. the lateralization group). Back-transformed data are plotted in Supplementary Figure B.

Figure 2:

Normalized ratio of lower trapezius/serratus anterior (LT/SA) muscle activity by group and motion (across intervals of motion). Log transformed data used for statistical analyses are plotted. The interaction of group and motion was statistically significant (p = 0.01), but pairwise comparisons of group in each motion were not. Error bars represent standard error. Log values greater than 0 represent a non-transformed LT/SA ratio greater than 1 (i.e., relatively more lower trapezius or less serratus anterior). Log values less than 0 represent a non-transformed LT/SA ratio less than 1 (i.e., relatively less lower trapezius or more serratus anterior).

A significant interaction of group and range of motion interval was found for the ratio of lower trapezius/anterior deltoid. For both groups, across all intervals, the mean transformed ratios were <0, indicating relatively less lower trapezius activity than anterior deltoid (Figure 3). The anterior tilt group had its highest ratio (i.e., the least negative transformed ratio) from 31–60° whereas the lateralization group had its lowest ratios from 31–90°. Pairwise comparisons using the Tukey HSD method did not reveal any significant meaningful comparisons of group and interval (e.g., lower trapezius/anterior deltoid activity from 31–60° in the anterior tilt group vs. the lateralization group). Back-transformed data are plotted in Supplementary Figure C. EMG magnitude interactions including group and group main effects were not statistically significant for the remaining muscles and ratios.

Figure 3:

Normalized ratio of lower trapezius/anterior deltoid (LT/AD) muscle activity by group and range of motion interval (across motions). Box-Cox transformed data used for statistical analyses are plotted. The interaction of group and interval was statistically significant (p = 0.03), but pairwise comparisons of group at each interval were not. Error bars represent standard error. Transformed values greater than 0 represent a non-transformed LT/AD ratio greater than 1 (i.e., relatively more lower trapezius or less anterior deltoid). Transformed values less than 0 represent a non-transformed LT/AD ratio less than 1 (i.e., relatively less lower trapezius or more anterior deltoid).

EMG Timing

Data were not normal and were heteroskedastic but were successfully transformed using a Box-Cox transformation. The three-way interaction of group, muscle, and motion was not significant (F_8,599 = 0.08, p = 1.00), nor were the two-way interactions of group and muscle (F_4,599 = 1.96, p = 0.10) or group and motion (F_2,599 = 0.11, p = 0.89). The two-way interaction of motion and muscle was significant (F_8,599 = 14.68, p < 0.01) (Figure 4). Back-transformed data are plotted in Supplementary Figure D. Tukey pairwise comparisons are provided in Table 4. Serratus anterior activation was significantly delayed approximately 7.5–13.5% range of motion in both groups compared to anterior deltoid and upper trapezius in all three motions and compared to middle deltoid in abduction. Lower trapezius activation was significantly delayed approximately 1.5–5% in both groups compared to anterior deltoid in flexion and the upper trapezius and middle deltoid in abduction. The main effect of group was not significant (F_1,43 = 1.54, p = 0.22). EMG timing of initiation means by muscle and motion are provided in Supplementary Table C.

Figure 4:

Muscle activity initiation by motion and muscle (across groups). Box-Cox transformed data used for statistical analyses are plotted. The interaction of motion and muscle was statistically significant (p < 0.01). Significant Tukey pairwise comparisons are denoted with *. Error bars represent standard error. LT = lower trapezius, SA = serratus anterior, UT = upper trapezius, AD = anterior deltoid, MD = middle deltoid.

Table 4:

Tukey pairwise comparisons of muscle initiation timing for the significant interaction of motion and muscle. Significant results (p < 0.05) are in bold and denoted with *. LT = lower trapezius, SA = serratus anterior, UT = upper trapezius, AD = anterior deltoid, MD = middle deltoid.

Comparison	Abduction	Flexion	Unrestricted Reach
LT vs. SA	t599 = −3.70, p = 0.02*	t599 = −3.12, p = 0.12	t599 = −4.39, p < 0.01*
LT vs. UT	t599 = 3.48, p = 0.04*	t599 = 1.26, p = 1.00	t599 = 2.39, p = 0.52
LT vs. AD	t599 = −2.22, p = 0.65	t599 = −3.65, p = 0.02*	t599 = −0.77, p = 1.00
LT vs. MD	t599 = 3.88, p = 0.01*	t599 = −6.24, p < 0.01*	t599 = −6.80, p < 0.01*
SA vs. UT	t599 = 7.15, p < 0.01*	t599 = 4.35, p < 0.01*	t599 = 6.76, p < 0.01*
SA vs. AD	t599 = −5.91, p < 0.01*	t599 = −6.77, p < 0.01*	t599 = −5.16, p < 0.01*
SA vs. MD	t599 = −7.57, p < 0.01*	t599 = 3.12, p = 0.11	t599 = 2.41, p = 0.51
UT vs. AD	t599 = 1.28, p = 0.99	t599 = −2.37, p = 0.54	t599 = 1.63, p = 0.95
UT vs. MD	t599 = −0.37, p = 1.00	t599 = 7.45, p < 0.01*	t599 = 9.15, p < 0.01*
AD vs. MD	t599 = 1.66, p = 0.94	t599 = −9.89, p < 0.01*	t599 = −7.56, p < 0.01*

Discussion

This study sought to determine if different muscle activity patterns exist in people with two specific scapular movement patterns. The hypothesis for EMG magnitude was that the anterior tilt group would have a higher average lower trapezius/serratus anterior muscle activity ratio than the lateralization group during arm elevation. The ratio was not consistently higher in the anterior tilt group, but the significant interaction of group and motion indicates that the two groups behaved differently across motions in terms of the ratio of lower trapezius/serratus anterior.

The ratio of lower trapezius/serratus anterior was greater than one in abduction for both groups and in flexion for the anterior tilt group, indicating relatively less serratus anterior activity than lower trapezius. Other studies (Briel et al., 2020, Michener et al., 2016) of asymptomatic participants found ratios of lower trapezius to serratus anterior that indicated relatively more serratus anterior muscle activity than lower trapezius. Additionally, in a sample of people with generalized shoulder pain, Michener et al. (2016) found lower ratios (ranging from 0.5–0.7) of lower trapezius/serratus anterior than the current study. Taken together, these findings suggest a possible relationship between these specific movement patterns (scapular anterior tilt and lateralization) and a relative reduction in serratus anterior muscle activity contribution to motion.

Interestingly, the possible relative reduction in serratus anterior muscle activity in these groups was more evident in the constrained motions (abduction and flexion) compared to the more natural, self-selected unrestricted reaching motion. Both groups had average lower trapezius/serratus anterior ratios less than one for unrestricted reaching. Ratios less than one indicate relatively more serratus anterior muscle activity than lower trapezius, which is more closely aligned with findings from asymptomatic and generalized shoulder pain participants found previously (Briel et al., 2020, Michener et al., 2016). Though speculative, this may suggest that individuals self-select movement patterns during functional motion using muscle activity patterns presumed to be more optimal. In this case, a more optimal muscle activity pattern may include increased serratus anterior activity as the serratus anterior is a main driver of scapular control both in terms of range of motion and stabilization on the thorax (Neumann and Camargo, 2019). Again although speculative, this could have implications for rehabilitation. For treatments in which the goal is to elicit more functional muscle activity patterns, it may be beneficial to instruct patients to use a self-selected movement pattern. Conversely, clinicians who are trying to identify minor deviations in muscle activity that may be clinically relevant may be more likely to elicit such deviations having patients perform planar motions (flexion/abduction).

Beyond reductions in serratus anterior muscle activity magnitude, findings of this study may indicate an additional reduction in lower trapezius muscle activity in the lateralization group early in the range of motion. The ratio of lower trapezius/anterior deltoid had a significant interaction of group and interval, and the ratio was lowest in the lateralization group from 31–90° of humerothoracic elevation (Figure 3). Anterior deltoid activity alone was lowest in the lateralization group during this interval of motion (Figure 1), so the reduced lower trapezius/anterior deltoid ratio was likely a result of a reduction in lower trapezius during that part of motion. Above 90°, the lateralization group increased its lower trapezius/anterior deltoid ratio while at the same time increasing anterior deltoid activity alone. This suggests that after 90°, the lateralization group had a relative increase in lower trapezius activity, potentially to compensate for less lower trapezius activity during early motion.

For EMG timing of initiation, no group effects were significant, so hypotheses based on group were not supported by this study’s findings. Rather, across both groups, there is evidence of possible delays in serratus anterior and lower trapezius activation. The EMG timing data indicate delays in the scapular muscles (lower trapezius, serratus anterior) compared to the other three muscles. Across both groups, lower trapezius activated later than upper trapezius in abduction, later than anterior deltoid in flexion, and later than middle deltoid in abduction. Serratus anterior activated later than upper trapezius and anterior deltoid in all three motions and later than middle deltoid in abduction. Studies of asymptomatic individuals (Wattanaprakornkul et al., 2011, Wickham et al., 2010) did not find statistically significant differences in initiation of lower trapezius, serratus anterior, and the other three tested muscles with the exception of earlier anterior deltoid initiation during flexion (Wattanaprakornkul et al., 2011). This suggests that the typical asymptomatic population would be expected to initiate muscle activation of the muscles simultaneously, and as such, the specific movement patterns in this investigation appear to have delayed serratus anterior and lower trapezius activity. Although group differences were not significant, delays in lower trapezius seem more profound in the lateralization group compared to the anterior tilt group (Supplementary Table C).

Limitations

This study had several limitations that should be considered when interpreting findings. The sample size for this study was originally determined using the lower trapezius/serratus anterior muscle activity ratio, yet post hoc power analyses indicated that the power to detect meaningful group differences for some comparisons was limited due to the variability within this study’s samples as compared to a priori projections. For example, the mean group difference in the lower trapezius/serratus anterior ratio in flexion was 0.3 (exceeding the meaningful group difference definition of 0.25), yet the post hoc power for this “worst case” analysis was 12%. However, despite this limited power for some comparisons, interactions of group and motion and group and interval were still found for anterior deltoid, lower trapezius/serratus anterior, and lower trapezius/anterior deltoid. Although pairwise comparisons did not achieve statistical significance, these interactions suggest that the two movement pattern groups behaved differently in terms of muscle activity patterns across range of motion or motion type.

Normalization of EMG signals is a known challenge with several proposed methods that each have advantages and disadvantages. EMG is most commonly normalized to MVCs, as done in this study. These MVCs are often isometric, which means that if the studied task is dynamic, the maximal contraction occurs with the muscle in a different length and electrodes in a different position relative to muscle fibers which affect signal intensity. Others have identified challenges with obtaining maximum contractions due to pain potentially limiting individuals’ ability to achieve a true maximum (Cooper and Karduna, 2022, Ettinger et al., 2016). Another challenge with obtaining maximum contraction references is providing sufficient resistance to actually obtain the maximum. Although steps were taken in this study to maximize examiner mechanical advantage, it is possible that maximum contractions were not always achieved due to insufficient resistance. While there are limitations to the MVC methods used in this study, they are similar to methods used in the majority of EMG studies, facilitating comparison to other studies. Additionally, the same method was applied to all participants in this study, so this ultimately should not have affected group EMG comparisons.

It is also possible that the movement pattern definitions used in this study were not extreme enough for clear EMG patterns to appear. While more exaggerated kinematic cutoffs might have revealed greater separation in group means for EMG, the prevalence of such patterns in the typical clinical population would be reduced, limiting the utility of these movement patterns as diagnostic classifications. A sub-analysis was performed to get an idea of the effect of stricter movement pattern definitions on EMG and is presented in Supplementary Materials. Group differences in EMG were not larger in the more distinct subgroups than they were for the entire participant sample, so the group movement pattern definitions and thresholds were likely not the reason why group EMG patterns were minimally different in this study.

Though this study found some EMG group differences in the form of significant interactions including group, it is possible that factors other than muscle activity of the five tested muscles contributed to the movement patterns of interest. Anterior tilt can be produced by active contraction of muscles attaching onto the anterior scapula (i.e., pectoralis minor, coracobrachialis, and short head of the biceps brachii) (Mohammed et al., 2016). The anterior tilt movement pattern could also be the result of resistance to posterior tilting despite activation of muscles that posteriorly tilt the scapula. This most likely occurs from shortness and passive resistance to stretch of the pectoralis minor, shortness of which has been demonstrated to occur alongside increased anterior tilt during motion (Borstad and Ludewig, 2005). The lateralization movement pattern could be the result of reduced muscle activity of other muscles that prevent lateral translation of the scapula (i.e., middle trapezius, rhomboid major, and rhomboid minor). Lateralization could also be the result of relatively increased activation of the muscles connecting the scapula to the humerus (the rotator cuff muscles in particular) compared to the scapular adductors. Shortness of these muscles could also cause the scapula to be pulled laterally during motion, as could tightness of the posterior glenohumeral capsule which has been demonstrated in cadavers (Mihata et al., 2020). As such, this study’s group categorization may have been an oversimplification of the processes involved in producing motion, and a more comprehensive assessment of additional factors like passive forces and anatomy (e.g., thorax shape) may be warranted.

Finally, this study lacked a control group of asymptomatic individuals without either movement pattern for statistical comparison. However, shoulder muscle activity of asymptomatic individuals has been studied previously (Briel et al., 2020, Michener et al., 2016, Wattanaprakornkul et al., 2011, Wickham et al., 2010), allowing for descriptive comparisons with the groups in this study. It is also worth noting that participants in this study had relatively high function and low symptom severity. The participants had DASH scores of about 12–13 on average. Previous studies on individuals with shoulder conditions seeking surgical intervention had an average preoperative DASH of 35 (Gummesson et al., 2003), and people with “rotator cuff lesions with or without rupture” who were not seeking surgical intervention had an average of 49.3 (Dupeyron et al., 2010). Pain ratings (Table 2) indicate that participants generally experienced shoulder pain they perceived to range from very mild to moderate. Future studies into these movement patterns would benefit from the inclusion of an asymptomatic kinematic control group and symptomatic participants with greater pain/disability for full statistical analyses.

Conclusion

Individuals with two scapular movement patterns had muscle activity magnitude differences in lower trapezius/serratus anterior, lower trapezius/anterior deltoid, and anterior deltoid. The anterior tilt group had a ratio of lower trapezius/serratus anterior >1 (indicating relatively more lower trapezius activity than serratus anterior) in two of three motions (abduction and flexion). The lateralization group had a lower trapezius/serratus anterior ratio >1 in one motion (abduction). Both groups had relatively increased serratus anterior muscle activity compared to lower trapezius (ratio <1) in the self-selected unrestricted reaching task. Early in the range of motion, the smaller lower trapezius/anterior deltoid ratio combined with the lower anterior deltoid activity alone in the lateralization group suggest a relative reduction in lower trapezius muscle activity in the lateralization group. For muscle activity timing of initiation, serratus anterior activation was significantly delayed approximately 7.5–13.5% range of motion in both groups compared to anterior deltoid and upper trapezius in all three motions and compared to middle deltoid in abduction. Lower trapezius activation was significantly delayed approximately 1.5–5% in both groups compared to anterior deltoid in flexion and the upper trapezius and middle deltoid in abduction. Future studies can use these findings to develop rehabilitation interventions based on patient-specific movement patterns and evaluate their effect on movement quality and symptoms.