The Effect of Scapular Orientation on Measures of Rotator Cuff Tendon Impingement: A Simulation Study

Abstract

Mechanical impingement of the rotator cuff tendons against the acromion (subacromial) and glenoid (internal) during shoulder motions has long been thought to contribute to tears. Clinically, the risk for impingement is thought to be influenced by scapular movement impairments. Therefore, our purpose was to determine the extent to which simulated changes in scapular orientation impact the proximity between the rotator cuff tendon footprint and the acromion and glenoid during scapular plane abduction. Specifically, shoulder kinematics were tracked in 25 participants using a high-speed biplane videoradiography system. Scapular movement impairments were simulated by rotating each participant’s scapula from their in vivo orientation about the scapular axes (±2°, ±5°, and ±10°). Subacromial and internal proximities were described using minimum distances, proximity center locations, and prevalence of contact. Statistical parametric mapping was used to investigate the extent to which these measures were impacted by simulated changes in scapular orientation. Simulated changes in scapular orientation significantly altered proximity patterns in a complex manner that depended on the impingement mechanism, humerothoracic elevation angle, and magnitude of the simulated change. Clinicians should be mindful of these factors when interpreting the potential effects during a clinical examination.

Mechanical impingement of the rotator cuff tendons by surrounding anatomy during shoulder motion has long been thought to contribute to tendon tears. The 2 primary impingement mechanisms involve deformation of the tendon bursal surface by the coracoacromial arch (subacromial impingement)¹ and of the articular surface against the superior glenoid rim (internal impingement).² Recent evidence suggests that subacromial impingement occurs during arm raising in about 50% of individuals, typically below 70° humerothoracic elevation.^3,4 By contrast, internal impingement, which was first identified in the abduction and external rotation positions, common to overhead athletics,^2,5 appears to occur in nearly all individuals when the humerus is positioned overhead.^6–8

Clinically, physical therapists are often interested in identifying the factors that underlie a patient’s condition to target interventions.⁹ Impaired scapular movement is often observed during a clinical exam¹⁰ and has been hypothesized to both contribute to and compensate for tendon impingement.^7,11–18 For example, insufficient upward rotation is hypothesized to position the acromion and glenoid closer to the rotator cuff tendons increasing the risk for impingement during shoulder motion.^7,11 Decreased scapular posterior tilt and external rotation have also been considered potentially problematic movement impairments.^12,13

Although these movement questions are fundamental to rehabilitation efficacy, there is a paucity of research investigating the consequence of movement impairments on mechanisms of pathology. Previous work has been limited to cadaveric simulations,^16,17,19 2-dimensional (2D) measures of tendon clearance that may be prone to error,²⁰ and between-subject comparisons that may be confounded by other factors (eg, anatomical differences).^14,15 While these studies have provided valuable preliminary evidence, a more comprehensive 3D analysis is warranted. Therefore, the purpose of this study was to determine the extent to which simulated changes made in scapular orientation impact the proximity between the rotator cuff tendon footprint and the acromion and glenoid during scapular plane abduction. We hypothesized that changes into scapulothoracic upward rotation, posterior tilt, and external rotation would significantly increase the minimum distance and do so in a dose–response manner.

Methods

Participants

Twenty-five participants (age: 55 [4] y, 64% female) were enrolled in this study as part of a larger investigation.²¹ Eligible participants were 50–60 years old without shoulder pain within the last 12 weeks. Participants were excluded if they currently smoked or had diabetes, a body mass index >32 kg/m², a history of shoulder symptoms following trauma, significant radiation exposure (eg, radiation therapy), or shoulder corticosteroid injection(s), subluxation, dislocation, adhesive capsulitis, fracture, osteoarthritis, or surgery. Participants enrolled in the parent study who achieved ≥150° humerothoracic elevation during motion capture procedures were included in the analysis. The study was approved by Henry Ford Health’s Institutional Review Board and complies with the Declaration of Helsinki. All participants provided written informed consent prior to data collection.

Data Collection

In vivo kinematics were quantified using a high-speed biplane videoradiography system.²² Participants were seated and asked to perform unloaded scapular plane abduction with their dominant shoulder. Practice trials were performed until the participant consistently demonstrated the correct movement plane and pace. Two 2-second trials were collected, but only the trial with the larger range of motion was selected for analysis. Additionally, participants underwent computed tomography (CT) scanning of their dominant shoulder. A standardized shoulder diagnostic ultrasound examination was also performed and subsequently interpreted by a fellowship-trained musculoskeletal radiologist to screen for shoulder pathology. During the examination, rotator cuff tendon thickness was estimated using previously described methods.²¹

Data Processing

CT images were processed using Mimics (Materialise NV) to create segmented volumes of the humerus, scapula, and third rib, which were used to track shoulder kinematics from the biplane images,²³ and to render 3D bone models for kinematic modeling. Once the bone models were created, anatomical coordinate systems were constructed by digitizing landmarks.²⁴ Notably, the scapular coordinate system was defined using a glenoid-oriented approach, which has been shown to correspond to the 3D axis of scapular motion during scapular plane abduction.²⁵ The selection of this coordinate system was important to ensure the simulated changes in scapular orientation (described below) were performed about physiologically meaningful axes. Compared with the scapular coordinate system recommended by the International Society of Biomechanics,²⁴ this glenoid-oriented coordinate system was generally oriented more anteriorly, inferiorly, and laterally, with an average transformation described by other authors.²⁶

Scapulothoracic kinematics were described using the Y-X′-Z″ rotation sequence as internal/external rotation (IR/ER), downward/upward rotation (DR/UR), and posterior/anterior tilt (PT/AT)²⁴ (Figure 1). Finally, each participant’s in vivo motion trial was reconstructed by animating their 3D bone models with their kinematics using custom software.²⁷

Figure 1 —

Changes in scapular kinematics were simulated by rotating each participant’s scapula from their in vivo orientation about the scapular axes, pivoting around a point located at the center of the humeral head. The scapular axes were defined using a glenoid-oriented coordinate system using the landmarks TS, AI, and GC.²⁵ Eighteen total simulations were performed in positive and negative directions about each of the 3 scapular axes across 3 magnitudes (2°, 5°, and 10°). The figure depicts a ±10° change in scapular orientation about each axis. AI indicates angulus inferior; GC, glenoid center; TS, trigonum scapulae.

To investigate the extent to which scapular kinematics impact subacromial and internal proximities, simulated changes in scapular kinematics were imposed by rotating each participant’s scapula from their in vivo orientation about the scapular axes, pivoting around a point located at the center of the humeral head (Figure 1). Eighteen total simulations were performed in positive and negative directions about each of the 3 scapular axes across 3 magnitudes (2°, 5°, and 10°). These magnitudes were selected to represent small, moderate, and large changes in scapular orientation reported following exercise interventions.^28–31 Although in vivo motion typically occurs concurrently about all 3 axes, it was decided to first investigate simple uniaxial effects before proceeding into more complex, multiaxial effects in future work.

Proximity patterns were described during the in vivo motion trial and each simulation by calculating distance maps between potential impinging structures (ie, acromion and glenoid) and the rotator cuff tendon footprint, defined as the superior and middle facet on the humeral greater tuberosity³² (Figure 2A). From these distance maps, 3 outcome measures were generated. First, the minimum distance was identified between the footprint and each potential impinging structure. For subacromial proximities, the identification of the minimum distance was constrained to along the articular margin region of the footprint where tendon thickness was estimated, which is relevant for the estimation of contact (described below) and corresponds to where rotator cuff tears are believed to originate.^33,34 For internal proximities, a constraint was imposed such that the minimum distance vector could not penetrate the humeral surface, which often occurs at lower humerothoracic elevation angles.⁶

Figure 2 —

Rotator cuff tendon insertion definitions and proximity calculations. (A) Superior view of the humeral head showing the articular margin area (dark gray) of the rotator cuff footprint (dark and light gray), defined as the superior and middle facets of the greater tuberosity. (B) Subacromial proximities were calculated between the acromion (semitransparent) and rotator cuff footprint (color-mapped). The location of the closest distance (black line) was constrained to occur along the articular margin of the footprint, corresponding to where the tendon thickness was estimated and where rotator cuff tears are believed to originate. (C) Internal proximities were calculated between the glenoid and rotator cuff footprint (color-mapped). The location of the closest distance (black line in B) is not visible given the minimum distance is very small. See color version online.

Second, the prevalence of tendon impingement was estimated by identifying potential contact between structures. Subacromial (acromion-to-footprint) contact was operationally defined to occur when the minimum distance was less than the subject-specific tendon thickness.¹⁴ Internal (glenoid-to-footprint) contact was operationally defined to occur when the minimum distance was less than the mean labral thickness reported by a prior study (4.3 mm).³⁵ A standardized value was needed given the labrum was not well visualized in the CT images used for the scapular reconstruction.

Third, the location on the tendon footprint that was closest to the acromion or glenoid (ie, proximity center) was identified using a weighted centroid method.³⁶ The proximity center was subsequently described relative to the humeral coordinate system and normalized to the diameter of a sphere fit to the humeral articular surface, to facilitate between-subject aggregation, and clinical interpretation.⁶

Statistical Analysis

The impact of simulated changes in scapular orientation on subacromial and internal proximities during scapular plane abduction was assessed using statistical parametric mapping (SPM). Based on findings of previous investigations,^4,6,7,14 the analyses for subacromial proximities were limited to 30° to 90° humerothoracic elevation while the analysis for internal proximities was limited to 90° to 150° humerothoracic elevation. One-factor repeated measures SPM analyses of variance (ANOVAs) were used to test whether significant differences existed in the proximity variable between the 7 levels of the independent variable (in vivo and ±2°, ±5°, and ±10° changes in scapular orientation). In the case of a significant ANOVA, post hoc SPM 2-tailed paired t tests were employed to compare each simulation with the in vivo condition using a Bonferroni correction (P < .0085, given 6 comparisons within each scapular orientation). For each analysis, changes in the proximity measure are reported as mean ± 95% confidence interval (CI). Full SPM results are presented in the Supplementary Materials ([available online]; see Figures S1–S18). Effect sizes (Cohen d) were also calculated to determine the practical significance of the effect and were interpreted using the thresholds suggested by Cohen (small: 0.20, medium: 0.50, and large: 0.80).³⁷ Finally, in participants with presumed contact, the angle at which contact first occurred was compared across simulations using 1-factor repeated measures ANOVAs. The proportion of participants in whom contact was alleviated by the simulated scapular movement was reported descriptively. SPM analyses were performed using the MATLAB version of the open source spm1d code. All other analyses were performed using SAS OnDemand for Academics (SAS Institute).

Results

On average, during scapular plane abduction, participants exhibited progressive scapulothoracic UR with comparatively minimal PT and IR/ER (Figure 3A). Early in the motion, the acromion-to-footprint minimum distance decreased and reached a minimum at an average (±95% CI) humerothoracic elevation angle of 44.8° (5.7°) before increasing for the remainder of the motion (Figure 3B). The average supraspinatus tendon thickness was 5.0 (0.2) mm. Of the 25 participants, 10 (40.0%) were presumed to experience subacromial contact beginning at an average humerothoracic elevation angle of 25.6° (5.1°) and ending at 67.0° (11.2°). The proximity center was located on the anteromedial aspect of the rotator cuff tendon footprint and shifted more anteriorly and laterally as the humerus elevated (Figure 3C).

Figure 3 —

Descriptive data (mean ± 95% confidence interval) for kinematics and proximity outcome variables during in vivo scapular plane abduction. (A) Scapulothoracic kinematics, (B) subacromial (acromion-to-footprint) and internal (glenoid-to-footprint) minimum distances and the ranges of motion in which presumed contact occurred, and (C) proximity center locations on the rotator cuff footprint. Scapulothoracic kinematics were described using a glenoid-oriented coordinate system, and upward rotation and anterior tilt were transformed to positive values. Proximity center locations are superimposed over an approximately average-sized humeral head (plot axes constrained to show only the anterior/lateral aspect of the humeral head, including the biceps groove). Glenoid-to-footprint minimum distance data are not available below 70° humerothoracic elevation as the minimum distance vector was constrained to not penetrate the humeral head, which occurred in most individuals below this angle.

By contrast, the glenoid-to-footprint minimum distances progressively decreased as the humerus elevated (Figure 3B). Contact was presumed to occur in 100% of participants beginning at an average humerothoracic elevation angle of 114.1° (4.3°) and continuing until maximum humerothoracic elevation (156.5° [2.0°]). The proximity center was located on the medial aspect of the rotator cuff tendon footprint and shifted anteriorly as the humerus elevated (Figure 3C).

Simulated changes in scapular orientation significantly impacted acromion-to-footprint minimum distances (Figure 4, see Figures S1–S3 and Table S1 in Supplementary Materials [available online]). Across all scapular motions, changes into UR/DR had the largest effect on the minimum distances. Specifically, when the arm was below approximately 40° humerothoracic elevation, changes into UR increased the minimum distance (2°: 0.1–0.2 mm, P = .006; 5°: 0.3–0.4 mm, P = .005; 10°: 0.4–1.1 mm, P = .003) while changes into DR decreased the minimum distances (2°: 0.1 mm, P = .006; 5°: 0.3 mm, P = .008; 10°: 0.2–0.4 mm, P = .008; Figure 4B). However, the effect reversed, such that by 55° humerothoracic elevation changes into DR increased the minimum distance (2°: 0.1–0.3 mm, P < .001; 5°: 0.3–0.8 mm, P < .001; 10°: 0.5–1.8 mm, P < .001), while changes into UR decreased the minimum distance (2°: 0.1–0.3 mm, P < .001; 5°: 0.3–0.8 mm, P < .001; 10°: 0.6–1.4 mm, P < .001). Simulated changes into ER/IR (Figure 4A) and AT/PT (Figure 4C) largely mirrored these findings but with smaller effect (average change ≤ 0.6 mm).

Figure 4 —

The effect of changing scapular orientation on the acromion-to-footprint (subacromial) minimum distance. (A) IR/ER, (B) DR/UR, and (C) PT/AT. Data are presented as the mean ± 95% confidence interval of the change from the in vivo condition. Asterisks (*) indicate that the minimum distance during the simulated change in scapular orientation is significantly different than the in vivo condition using statistical parametric mapping-repeated measures ANOVA and post hoc testing (2-tailed paired t tests, P < .0085). ANOVA indicates analysis of variance; AT, anterior tilt; DR, downward rotation; ER, external rotation; IR, internal rotation; PT, posterior tilt; UR, upward rotation.

Most simulated changes in scapular orientation significantly impacted the normalized proximity center location on the rotator cuff tendon footprint and generally did so below 60° humerothoracic elevation (Figure 5, see Figures S4–S9 in Supplementary Materials [available online]). Specifically, changes into ER shifted the proximity center location medially (2°: 0.6%, P = .001; 5°: 1.1%–1.4%, P < .001; 10°: 1.3%–2.9%, P < .001) but had no effect on the anterior/posterior position (SPM ANOVA: P = .050, all SPM post hoc: P = 1.0; Figure 5A). Furthermore, changes into UR shifted the proximity center location posterior (2°: 0.8%–1.2%, P < .001; 5°: 1.8%–2.9%, P < .001; 10°: 2.3%–6.0%, P < .001) and medially (2°: 0.7%–1.0%, P < .001; 5°: 1.1%–2.3%, P < .001; 10°: 2.7%–4.3%, P < .001; Figure 5B), while changes into PT shifted the proximity center posterior (2°: ≤1.0%, P = .002; 5°: 1.2%–1.8%, P < .001; 10°: 2.8%–3.5%, P < .001) and laterally (2°: 0.5%–1.1%, P < .010; 5°: 0.9%–2.2%, P < .001; 10°: 1.7%–4.2%, P < .001; Figure 5C). Simulated changes into IR, DR, and AT largely mirrored these findings except the shift on the footprint occurred in the opposite direction.

Figure 5 —

The effect of changing scapular orientation on the normalized proximity center location on the rotator cuff footprint for acromion-to-footprint (subacromial) proximities. (A) IR/ER, (B) DR/UR, and (C) PT/AT. The solid lines represent the trajectories of the mean change in proximity center location between each simulation and the in vivo condition. Error bars are used to represent the variability (95% confidence intervals) of the change at 30° and 90° humerothoracic elevation. Data are superimposed over an approximately average-sized humeral head (plot axes constrained to show only the anterior/lateral aspect of the humeral head, including the biceps groove). AT indicates anterior tilt; DR, downward rotation; ER, external rotation; IR, internal rotation; PT, posterior tilt; UR, upward rotation.

Across all scapular motions, increasing UR alleviated acromion-to-footprint contact in the highest proportion of participants (80%, regardless of the simulation magnitude; Table 1). Furthermore, a 10° change into UR shifted the angle of contact higher into humerothoracic elevation (13.0° [4.5°] change, P < .001). By contrast, DR alleviated contact in 0% of individuals and tended to shift the angle of contact to lower humerothoracic elevation, but the effect did not meet statistical significance (2.0°–4.5° change, P ≥ 0.401). ER (2°: 20%, 5°: 70%, and 10°: 80%), PT (2°: 30%, 5°: 40%, and 10°: 40%), and AT (2°: 20%, 5°: 50%, and 10°: 60%) also alleviated contact, but did not significantly impact the angle of initial contact.

Table 1.

The Prevalence of Presumed Acromion-to-Footprint (Subacromial) Contact Under in Vivo Conditions and Each Simulated Change in Scapular Orientation

	Contact N (%)	Alleviated contact N (%)	Angle of initial contact (mean ± 95% CI)	P
In vivo	10 (40%)	N/A	25.6° ± 5.1°	N/A
Internal/extemal rotation				<.001
Internal rotation
2°	10 (100%)	0 (0%)	24.9° ± 4.9°	.993
5°	10 (100%)	0 (0%)	24.9° ± 4.9°	.260
10°	10 (100%)	0 (0%)	23.8° ± 4.9°	.001
External rotation
2°	8 (80%)	2 (20%)	26.0° ± 4.9°	.857
5°	3 (30%)	7 (70%)	26.8° ± 4.9°	.799
10°	2 (20%)	8 (80%)	28.0° ± 4.9°	.024
Downward/upward rotation				<.001
Downward rotation
2°	10 (100%)	0 (0%)	27.0° ± 7.3°	.974
5°	10 (100%)	0 (0%)	25.9° ± 7.3°	.162
10°	10 (100%)	0 (0%)	24.6° ± 7.1°	<.001
Upward rotation
2°	2 (20%)	8 (80%)	31.0° ± 7.1°	.968
5°	2 (20%)	8 (80%)	34.7° ± 7.3°	.784
10°	2 (20%)	8 (80%)	42.1° ± 7.3°	.401
Posterior/anterior tilt				.769
Posterior tilt
2°	7 (70%)	3 (30%)	24.5° ± 4.7°	NT
5°	6 (60%)	4 (40%)	24.3° ± 4.7°	NT
10°	6 (60%)	4 (40%)	24.6° ± 4.9°	NT
Anterior tilt
2°	8 (80%)	2 (20%)	25.2° ± 4.7°	NT
5°	5 (50%)	5 (50%)	26.5° ± 4.7°	NT
10°	4 (40%)	6 (60%)	25.7° ± 4.7°	NT

Abbreviations: CI, confidence interval; N/A, not applicable; NT, not tested (due to insignificant 1-factor repeated-measures analysis of variance).

Simulated changes in scapular orientation significantly impacted glenoid-to-footprint minimum distances during scapular plane abduction (Figure 6, see Figures S12–S14 and Table S2 in Supplementary Materials [available online]). Across all scapular orientations, changes into UR had the largest effect on increasing the minimum distances (2°: 0.2–1.3 mm, P < .001; 5°: 0.6–3.4 mm, P < .001; 10°: 1.4–7.3 mm, P < .001; Figure 6A) followed by ER (2°: 0.1–0.8 mm, P < .001; 5°: 0.2–2.1 mm, P < .001; 10°: 0.6–3.8 mm, P < .001; Figure 6B) and AT (2°: 0.1–0.3 mm, P < .001; 5°: 0.2–0.5 mm, P < .001; 10°: 0.3–1.0 mm, P < .001; Figure 6C). Simulated changes into DR, IR, and PT largely mirrored these findings except the simulated change in scapulothoracic orientation decreased in the minimum distance.

Figure 6 —

The effect of changing scapular orientation on the glenoid-to-footprint (internal) minimum distance. (A) IR/ER, (B) DR/UR, and (C) PT/AT. Data are presented as the mean ± 95% confidence interval of the change from the in vivo condition. Asterisks (*) indicate that the minimum distance during the simulated change in scapular orientation is significantly different than the in vivo condition using statistical parametric mapping repeated measures ANOVA and post hoc testing (2-tailed paired t tests, P < .0085). ANOVA indicates analysis of variance; AT, anterior tilt; DR, downward rotation; ER, external rotation; IR, internal rotation; PT, posterior tilt; UR, upward rotation.

Changes in scapular orientation significantly impacted the normalized proximity center location on the rotator cuff tendon footprint in a complex manner (Figure 7, see Figures S15–S18 in Supplementary Materials [available online]). Simulated changes into ER shifted the proximity center location anteriorly (2°: 0.7%–1.4%, P < .001; 5°: 1.7%–4.0%, P < .001; 10°: 3.1%–13.3%, P < .001), and medially (2°: 0.6%–0.7%, P < .001; 5°: 1.2%–1.7%, P < .004; 10°: 2.7%–3.1%, P < .001) when the arm was above approximately 100° humerothoracic elevation. Furthermore, simulated changes of 5° to 10° into DR shifted the proximity center anteriorly when the arm was above 100° humerothoracic elevation (5°: 0.2%–0.3%, P < .001; 10°: 0.3%–0.8%, P < .010) and laterally above 130° humerothoracic elevation (5°: 0.6%–1.4%, P < .004; 10°: 1.5%–3.8%, P < .001; Figure 7B). Simulated changes in AT/PT had little effect on the normalized proximity center location (Figure 7C).

Figure 7 —

The effect of changing scapular orientation on the normalized proximity center location on the rotator cuff footprint for glenoid-to-footprint (internal) proximities. (A) IR/ER, (B) DR/UR, and (C) PT/AT. The solid lines represent the trajectories of the mean change in proximity center location between each simulation and the in vivo condition. Error bars are used to represent the variability (95% confidence intervals) of the change at 90° and 150° humerothoracic elevation. Data are superimposed over an approximately average humeral head size (plot axes constrained to show only the anterior/lateral aspect of the humeral head, including the biceps groove). ANOVA indicates analysis of variance; AT, anterior tilt; DR, downward rotation; ER, external rotation; IR, internal rotation; PT, posterior tilt; UR, upward rotation.

Across all scapular motions, increasing UR alleviated glenoid-to-footprint contact in the highest proportion of participants. Specifically, a simulated change into UR of any magnitude resulted in alleviated contact in 100% of participants (Table 2). Furthermore, a simulated change into UR delayed initial contact to higher humerothoracic elevation angles (2°: 3.6° [1.8°] change, P = .001; 5°: 9.6° [1.8°] change, P < .001; 10°: 20.0° [1.8°] change; P < .001). By contrast, DR alleviated contact in 0% of individual and tended to shift the angle of contact to lower humerothoracic elevation angles (2°: 3.4° [1.8°] change, P = .004; 5°: 8.3° [1.8°] change, P < .001; 10°: 16.0° [1.8°] change; P < .001). External rotation (2°: 80%, 5°: 88%, 10°: 96%), and AT (2°: 64%, 5°: 92%, 10°: 92%) also alleviated contact with modest (≤8.2°) changes in the angle of initial contact. IR and PT alleviated contact in the fewest participants (<8%).

Table 2.

The Prevalence of Presumed Glenoid-to-Footprint (Internal) Contact Under in Vivo Conditions and Each Simulated Change in Scapular Orientation

	Contact N (%)	Alleviated contact N (%)	Angle of initial contact (mean ± 95% CI)	P
In vivo	25 (100%)	N/A	114.1° ± 4.3°	NA
Intemal/extemal rotation				<.001
Internal rotation
2°	24 (96%)	1 (4%)	112.5° ± 4.5°	.739
5°	23 (92%)	2 (8%)	110.1° ± 4.5°	.003
10°	24 (96%)	1 (4%)	105.5° ± 4.5°	<.001
External rotation
2°	5 (20%)	20 (80%)	115.9° ± 4.5°	.590
5°	3 (12%)	22 (88%)	118.0° ± 4.5°	.004
10°	1 (4%)	24 (96%)	122.3° ± 4.5°	<.001
Downward/upward rotation				<.001
Downward rotation
2°	25 (100%)	0 (0%)	110.7° ± 4.3°	.001
5°	25 (100%)	0 (0%)	105.7° ± 4.3°	<.001
10°	25 (100%)	0 (0%)	98.1° ± 4.3°	<.001
Upward rotation
2°	0 (0%)	25 (100%)	117.7° ± 4.3°	.001
5°	0 (0%)	25 (100%)	123.7° ± 4.3°	<.001
10°	0 (0%)	25 (100%)	117.7° ± 4.3°	<.001
Posterior/anterior tilt				<.001
Posterior tilt
2°	25 (100%)	0 (0%)	113.3° ± 4.3°	.539
5°	25 (100%)	0 (0%)	111.6° ± 4.3°	<.001
10°	25 (100%)	0 (0%)	109.4° ± 4.3°	<.001
Anterior tilt
2°	9 (36%)	16 (64%)	115.0° ± 4.3°	.324
5°	2 (8%)	23 (92%)	116.5° ± 4.3°	<.001
10°	2 (8%)	23 (92%)	118.3° ± 4.3°	<.001

Abbreviations: CI, confidence interval; N/A, not applicable; NT, not tested (due to insignificant 1-factor repeated-measures analysis of variance).

Discussion

This study aimed to determine the extent to which simulated changes in scapular orientation impact the proximity between the rotator cuff tendon footprint, and the acromion, and glenoid. These simulations were imposed over in vivo data that largely reflected what has been described previously.^3,4,6–8 For subacromial proximities, the rotator cuff tendon footprint was closest to the acromion below 70° humerothoracic elevation with presumed in vivo contact occurring in <50% of individuals.^3,4 For internal proximities, the footprint progressively approached the glenoid above 90° humerothoracic elevation with presumed in vivo contact occurring in nearly all individuals.^6–8 When simulated changes in scapular orientation were imposed, proximity patterns were altered in a complex manner that depended on the impingement mechanism (subacromial, internal), humerothoracic elevation angle, and magnitude of the simulated change.

From the subacromial impingement perspective, simulated changes in scapular orientation altered the acromion-to-footprint minimum distance in a complex manner with the direction of effect depending on the humerothoracic elevation angle. At lower elevation angles (below ~40°), simulated changes into UR, ER, and AT increased the minimum distance while the opposite occurred at higher angles. This reversal of effect likely reflects the changing orientation between the rotator cuff footprint and acromial undersurface as the humerus abducts. For example, at lower elevation angles, the footprint and acromial surfaces are approximately parallel (Figure 2B). As humeral abduction occurs, however, the footprint rotates under the acromion making the structures more oblique to one another (Figure 2C). At first, this angulation brings the footprint closer to the acromion, but eventually the footprint “clears” the acromion and moves further away (Figure 3B). Accordingly, scapular orientation changes that decrease the minimum distance at lower humerothoracic elevation angles appear to increase the minimum distance at higher angles.

Regardless of the direction of effect, simulated changes in scapular orientation generally resulted in small effect sizes (see Table S1 in Supplementary Materials [available online]) and small changes in the minimum distance (<1.8 mm), despite reaching statistical significance. This magnitude of change is consistent with the findings of Seitz et al, who used ultrasound to measure changes in acromiohumeral distance during the scapular assistance test.²⁰ During the test, a patient raised their arm while a tester manually facilitated UR, ER, and PT, resulting in an average minimum distance change of ≤ 2.1 mm. Unfortunately, it remains unclear whether the small changes in subacromial minimum distance reported in the current study and by Seitz et al have clinical meaningfulness. Researchers have previously used a change of 2.0 mm as a threshold for meaningfulness.^14,20 However, this threshold is based on the minimal detectable change, which uses measurement reliability to guide decisions about potential clinical meaningfulness. Ideally, a threshold for clinical meaningfulness is one that directly reflects a relationship between changes in proximities and clinical measures (eg, pain and function), which should be a focus of future investigations.

From the internal impingement perspective, simulated changes in scapular orientation impacted glenoid-to-footprint minimum distances in a consistent, dose–response manner: UR, ER, and AT consistently increased the minimum distance, while DR, IR, and PT decreased the minimum distance. This dose–response pattern was also reflected in the effect sizes (see Table S2 in Supplementary Materials [available online]), which varied widely from small (|d| ≤ 0.20) to large (|d| ≥ 0.80). Simulated changes into UR and ER resulted in the largest increase in minimum distance (up to 7.3 mm) by rotating the posterosuperior glenoid away from the tendon footprint, with modest to large effect sizes occurring with as little as 5° changes into UR (see Table S2 in Supplementary Materials [available online]). This finding is interesting as it suggests that a 5° change in scapular orientation, which is often considered clinically unimportant,^18,29 may alter the risk for internal impingement during scapular plane abduction. Importantly, this magnitude of change has been reported following targeted scapulothoracic strengthening programs.^28,30 Although it remains unclear whether improvements in symptoms are related to changes in proximity measures, results of the current study suggest that large changes in kinematics may not be necessary to alter glenoid-to-footprint proximities.

This study also investigated the proximity center location on the footprint to further characterize the spatial relationships of tendon impingement. During the in vivo motion trial, the subacromial proximity center moved laterally across the footprint as the humerus elevated while the internal proximity center moved anterior along the medial aspect of the footprint (Figure 3C). Although most simulated changes in scapular orientation significantly altered these locations, changes that shifted the proximity center to the medial aspect of the footprint were of particular interest because degenerative tears are believed to originate in that region.^33,34 Changes into UR, ER, and AT shifted the subacromial proximity center medially while changes into ER shifted the internal proximity center medially.

These data on proximity center location offer additional information that may be helpful when interpreting bone-to-bone minimum distances, which have inherent limitations when used to investigate mechanisms of soft tissue pathology. Interestingly, UR, ER, and AT tended to move the acromion further away from the footprint yet positioned it nearer to the medical aspect of the tendon footprint, where rotator cuff pathology tends to occur.^33,34 These findings seem contradictory when interpreted within the context of potential mechanisms of rotator cuff pathology, and highlights the need for more complex investigations that incorporate direct measures of soft tissue deformation.

The results pertaining to the alleviation of presumed contact were perhaps the most informative regarding how simulated changes in scapular orientation alter subacromial and internal proximities. Regardless of mechanism, simulated changes as small as 2° into UR alleviated contact in >80% of individuals. Moreover, increasing UR essentially increased the shoulder’s hypothetical impingement-free functional workspace by delaying presumed glenoid-to-footprint contact. For example, a 10° change into UR shifted the angle of initial contact from 114° to 134° humerothoracic elevation. This 20° increase in the hypothetical functional workspace may be substantial for individuals whose functional and occupational tasks require overhead work that likely exposes them to internal impingement.

An important consideration when interpreting the results of this study is how they compare with previous work that investigated differences in proximities between subjects grouped by scapular classifications.^14,15 In these previous studies, differences in proximities measures were compared between 2 groups of individuals (low vs high UR), and were generally found to be small, or not statistically significant. By contrast, the current study investigated the effect of uniaxial changes in scapular orientation and reported moderate to large statistically significant effects, particularly for internal minimum distances. These disparate findings between studies are likely explained, in part, by differences in methodology. For example, although participants in the prior study were classified based on scapular UR, differences may also exist in other kinematic variables (eg, PT and glenohumeral position) that may have confounded the results. By contrast, the current study investigated uniaxial changes in scapular orientation within an individual. Although uniaxial changes in scapular kinematics likely do not reflect what occurs in vivo, the methodological approach allows for more controlled assessment of how changes in movement impact impingement mechanisms. The ability to interpret the consequences of uniaxial scapular rotations may have important clinical implications because movement modification tests (eg, scapular assistance test) typically involve facilitating movement about multiple axes simultaneously (eg, UR, PT, and ER)^20,38; however, it appears PT has an opposite effect on minimum distances compared with UR and ER. Thus, performing movement modification tests with a combination of movements may offset each other and confound clinical assessment of the relationship between scapular movement, and mechanisms of impingement, and symptom provocation.

The discrepancy in findings between the current study and prior work^14,15 also suggests that other factors may confound between-subject analyses that were inherently controlled for in the within-subject design utilized by the current study. For example, bony morphology, capsular tightness, and other morphological or adaptive characteristics may directly, or indirectly, alter subacromial and internal impingement proximities. Furthermore, we did not seek to distinguish the source of movement variability between our participants. In this way, the between-subject movement variability within our sample may have helped to enhance the generalizability of the results. In the prior study, it is possible that individuals classified as having low UR may have had a more superiorly oriented glenoid (ie, glenoid inclination), which negated the need for orienting the glenoid superiorly through UR. This potential interaction between morphology and kinematics is inherently controlled for in the current study given the within-subject analysis; however, more research is needed to investigate the interaction to understand the sources and implications of movement variability that is often observed between individuals. This information may help inform the development of movement-based diagnostic classifications.⁹

This study has limitations to consider when interpreting the results. First, changes in scapular orientation were simulated to occur about one scapular axis pivoting around the humeral head center, which is not likely what occurs in vivo when more complex movement impairments are often observed (eg, scapular dyskinesis).³⁹ Although this methodological decision is likely an over-simplification, it was intentional so that we can first understand simple effects before moving into more complex combinations in future work. Further, the multiplicative and sequence-dependent nature of rotation matrices makes it likely that small (typically < 1°) rotations about secondary axes may have also occurred. Second, some of the more extreme changes in scapular orientation may not be physiologically possible in some individuals due to constraints imposed by the thorax. However, during pilot testing we explored physiological plausibility by calculating the displacement of the scapular inferior angle (angulus inferior) and the root of scapular spine (trigonum scapulae) during simulated 10° changes in AT/PT and IR/ER, respectively. During these simulated motions, the inferior angle displaced an average of 2.0 cm and the root of the scapular spine displaced an average of 2.2 cm. Unfortunately, a full reconstructed model of the thorax was not included in our kinematic model for us to assess whether the scapula intersected with the thorax during the 10° simulated changes. However, large (10°) changes in scapular kinematics have been reported previously,^28,30 suggesting it may be possible when rotations occur about 3 axes simultaneously (a consideration acknowledged above). Third, the use of bone-to-bone proximity measures does not allow for direct assessment of tendon impingement. In particular, glenoid-to-footprint minimum distances were interpreted relative to a standardized labral thickness.^6,35 Although there is variability in labral dimensions across individuals, prior work established that the assumed thickness did not change the determination of contact in most individuals. Ultimately, more sophisticated modeling (eg, finite element) is needed to more comprehensively investigate the factors that influence impingement mechanisms. Finally, this study only investigated changes in scapular orientation during scapular plane abduction therefore the results may not reflect what may occur during other shoulder motions (eg, throwing). Taken together, the assumptions and limitations of this modeling study need to be carefully considered when interpreting the results.

In conclusion, simulated changes in scapular orientation impacted subacromial and internal proximity patterns in a complex manner that is unique to each impingement mechanism. The finding that altering scapular orientation impacts proximities differently across humerothoracic elevation angles is important because it suggests that clinicians should be mindful of where in the range of motion symptoms and scapular movement impairments are observed when interpreting the potential effects during a clinical examination. For example, pain at lower angles (below 70°) may suggest subacromial impingement,^3,4 pain near mid-range (~90°) may suggest contractile irritation as a result of higher muscle forces,^40,41 and pain at higher angles (above 90°) may suggest internal impingement.^6–8 Once a mechanism for a patient’s symptoms is hypothesized, clinicians can then carefully use the results of this study to assess the potential implications of any movement impairments observed during clinical examination, and if deemed appropriate, to help guide movement-based treatment strategies. Of course, this clinical decision-making process remains untested is therefore speculative and remains an important direction for future work to better understand the movement-based mechanisms of shoulder pain.