The interplay between capsuloligamentous restraints and muscles is not fully appreciated in the maintenance of stability at the glenohumeral joint, particularly in the anterior-posterior direction. The relative contributions of static and dynamic stabilizers vary throughout the range of motion as the joint capsule and ligaments may be loose or taut depending on position1-4. Some have suggested that preparatory and reactive muscle-recruitment strategies must also vary to compensate for these changes and to preserve joint integrity1-3,5,6. However, more research that simultaneously quantifies muscle recruitment and the level of glenohumeral stability at different positions is needed. The amount of joint stability that is achieved in response to external loads, or perturbations, can be described by measuring stiffness7-9. Understanding joint stability with respect to stiffness regulation strategies is clinically important, particularly in the shoulder, because of its high mobility, inherent laxity, and predisposition to acute and chronic injuries.
Stiffness is the resistance of a tissue to length and/or position changes10,11. It is a mechanical property that is determined by calculating the ratio of change in force per change in position, and it represents a fundamental measure of joint stability in response to load10-12. During shoulder movements, stiffness is regulated through preparatory and reactive muscle-recruitment strategies based on the functional demands of a task but balanced with the need for stability1-3,7-9,13. Joint position and laxity are factors that may influence stiffness regulation and could have clinical implications. Shoulder abduction and external rotation (the apprehension position) has been associated with an increased risk of injury, as has increased tissue laxity4,8,14-16.
Much of our knowledge about how muscles and ligaments respond to external forces has been derived from cadaver studies4,17,18. Studies that have quantified shoulder stiffness in vivo, especially studies that have replicated angular movements during functional loads and positions, have been limited. Several studies have examined passive stiffness in linear directions to determine "end feel" or to corroborate clinical laxity assessments19-22. Zhang et al.23 measured in vivo shoulder stiffness about an abduction axis and demonstrated increased stiffness with higher levels of torque production. Those findings support the theory that muscle activity enhances joint stability in the shoulder during movement. However, what remains unknown is the preparatory and reactive muscle-activation strategies employed to regulate rotational stiffness in "apprehensive" positions, which are commonly associated with complaints of shoulder instability.
The primary purpose of the present study was to determine whether stiffness regulation and muscle-activation strategies were different under varying joint positions and levels of rotational torque. A secondary purpose was to determine the relationship between generalized joint laxity, glenohumeral joint laxity, and joint stiffness. We hypothesized that shoulder stiffness is dependent on muscle activation and joint position, such that joint stability is enhanced with muscle activation but diminishes from neutral to an externally rotated position. We also theorized that a negative linear relationship would exist between the laxity measures and stiffness.
Research Design
A stratified post-test-only one-group design was used in the present study. The independent variables were joint position (90° of abduction with 0° of external rotation and 90° of abduction with 90% of maximum external rotation) and the level of internal rotation torque (0% and 50% of maximum internal rotation torque). The 0% torque level was selected to represent the contributions of passive tissues and resting muscle tone, whereas the 50% torque level was based on previous research demonstrating that stiffness remains relatively constant between 30% and 80% of maximum9. Pilot testing demonstrated that 50% internal rotation torque was the highest that subjects could reliably maintain. The dependent variables were shoulder stiffness (measured in Nm/deg); generalized joint laxity (as indicated by a Beighton score of 0 to 9); glenohumeral joint laxity in the anterior, posterior, and inferior directions (as indicated by a grade of 1 to 3); and the amplitude of the electromyographic signal before (preparatory) and after (reactive) the perturbation.
Subjects
Forty healthy, physically active individuals (including twenty male and twenty female subjects) with a mean age (and standard deviation) of 25.2 ± 4.6 years, a mean height of 174.7 ± 6.7 cm, and a mean mass of 73.1 ± 13.8 kg participated in the study. An individual was considered to be physically active if he or she performed a minimum of thirty minutes of exercise three or more times a week. The dominant arm, which was the arm used to throw a ball, was used for testing.
The exclusion criteria were participation in upper extremity sports within the previous three years, systemic disease, metabolic or neurological disorders, previous shoulder surgery, and current limitations due to a shoulder abnormality. Female subjects were only included if they were within Day 0 to 12 of their menstrual cycle (the follicular phase) to minimize the potential hormone influence on ligamentous laxity24-26. All subjects completed a physical activity and health history questionnaire. University institutional review board approval and informed consent were obtained prior to data collection.
Range-of-Motion Measurements
Glenohumeral internal and external rotation of both arms were measured with the subject supine with use of a goniometer (Medco Sports Medicine, Tonawanda, New York) and a bubble level. The orthopaedic surgeon applied a downward pressure on the acromion process to stabilize the scapula while the forearm was passively moved into internal or external rotation until a capsular end point was detected, and a second investigator (K.C.H.) measured the joint angle. Reliable goniometric measurements were observed by using the same investigator with intraclass correlation coefficients of 0.991 for external rotation and 0.988 for internal rotation.
Shoulder Stiffness Measurements
A stiffness-testing instrument was used to apply a standardized perturbation to the shoulder and to measure resistance torque (N-m). The instrument was also used to measure isometric peak internal rotation torque (N-m). The instrument consists of a steel frame, servomotor, gearbox, and attachment arm. A brushless servomotor (model #R67GENA-R2-NS-NV-00; Pacific Scientific, Rockford, Illinois) mated with a gearbox (model #AB180-S2-P2; Apex Dynamics, Holbrook, New York) was mounted in a 3-cm-thick steel plate that was offset from the main frame. A torque-reaction sensor (model #T5400; Futek Advanced Sensor Technology, Irvine, California) (maximum, 565 N-m) and adaptor arm were coupled to the gearbox below the steel plate. Signals from the torque-reaction sensor (analog output range, 0 to 10 VDC) were sent as an analog signal to a personal computer, where it was stored and analyzed with MyoResearch software (Noraxon USA, Scottsdale, Arizona).
Post hoc analyses of three repetitions for all forty subjects were performed on torque values to calculate shoulder stiffness and to establish intertest reliability. The intertest reliabilities for torque were 0.89 (with 90% of maximum external rotation and 0% internal rotation torque), 0.98 (with 90% of maximum external rotation and 50% of maximum internal rotation torque), 0.92 (with 0° of external rotation and 0% internal rotation torque), and 0.95 (with 0° of external rotation and 50% of maximum internal rotation torque).
Electromyographic Measurements
The contribution of muscle activity to shoulder stiffness was recorded during external rotation perturbations. Surface and fine-wire electromyography activity were simultaneously recorded from the supraspinatus, infraspinatus, teres minor, subscapularis, and anterior deltoid muscles27-31. Self-adhesive Ag/AgCl bipolar surface electrodes (#272; Noraxon USA) were applied to the skin over the anterior deltoid and ipsilateral clavicle (ground electrode) according to standard skin preparation and placement procedures32,33.
Intramuscular fine-wire bipolar electrodes were placed into the supraspinatus, infraspinatus, teres minor, and subscapularis under sterile conditions with use of the single-needle technique described by Basmajian and De Luca32. Dual 50.8-µm-diameter Stablohm 800A stainless steel wires insulated with heavy-poly nylon (California Fine Wire, Grover, California) were used according to the technique described by Basmajian and Stecko34. Electrode placement in the supraspinatus, infraspinatus, and teres minor was based on the report by Perotto35, and electrode placement in the subscapularis was based on the report by Kadaba et al36 (Fig. 1). Electrode placement was confirmed by monitoring electromyography activity during isolated muscle testing of each muscle and electromyography signal identification with use of MyoResearch software37.
Bipolar electrodes were connected to spring cables on a Noraxon TeleMyo System frequency-modulated (FM) electromyograph transmitter (Noraxon USA)38. A single-ended amplifier was used (impedance >10 MO; gain = 1000) with a fourth-order Butterworth filter (10 to 500 Hz) and a common-mode rejection ratio of 130 dB. A receiver with a sixth-order filter (gain = 2, total gain = 2000) was used to further amplify the signal. The signal was passed to a computer with a sixteen-channel, 12-bit analog-to-digital card (Keithley Metrabyte DAS-1000; Keithley Instruments, Taunton, Massachusetts). Data were sampled at a rate of 2500 Hz following the International Society of Electrophysiology and Kinesiology and previously established guidelines31,39,40. Raw electromyography data were processed and analyzed with MyoResearch software (version 2.11; Noraxon USA) with full-wave rectification (i.e., linear envelope detection), integrated with a sixth-order Butterworth filter, and smoothed over a 15-ms moving window. Muscle activity area was calculated as the sum of the amplitudes of integrated electromyography activity over the total time of the trial. Area was analyzed during two time-periods: preparatory (100 ms prior to the perturbation) and reactive (250 ms following the perturbation). The data were normalized to the maximum voluntary isometric contraction. Previous research has demonstrated moderate to high test-retest correlation coefficients for electromyography area measures, ranging from 0.80 to 0.9941.
Laxity Measurements
Generalized joint laxity was assessed with the Beighton mobility scale, which includes (1) at least 10° of elbow hyperextension, (2) at least 10° of knee hyperextension, (3) passive opposition of the thumb so that it is at least 0.5 cm from the volar aspect of the forearm, (4) at least 90° of passive hyperextension of the second to fifth digits, and (5) at least 45° of ankle dorsiflexion27. A point for each characteristic present, tested bilaterally, was assigned to the subject (total possible score, 9). A generalized joint laxity score of =4 (indicating the presence of four or more characteristics) was the criterion measure for generalized joint laxity42.
The load-and-shift and sulcus special tests were performed to assess anterior, posterior, and inferior glenohumeral joint laxity19,43-45. Grade 1 was defined as humeral head movement to the glenoid labrum rim. Grade 2 was defined as humeral head movement over the glenoid labrum rim that spontaneously reduced when load was removed. Grade 3 was defined as humeral head movement over the glenoid labrum rim that did not spontaneously reduce when load was removed46. A board-certified orthopaedic surgeon who was experienced in shoulder laxity assessments graded humeral head displacement in each direction as grade 1, 2, or 346. Applied loads were not standardized between subjects by the physician; however, previous research has demonstrated good to excellent intratester agreement (? = 0.84) and test-retest correlation coefficients (between 0.74 and 0.92)47,48.
Experimental Procedures
Laxity and range-of-motion assessments were performed first, followed by skin preparation and surface electrode application. The subject was then positioned prone, and the physician inserted fine-wire electrodes into the supraspinatus, infraspinatus, teres minor, and subscapularis. After the electrodes had been placed, the subject performed a warm-up consisting of stretching as well as active and resisted range of motion for shoulder flexion, extension, abduction, external rotation, and internal rotation.
The subject was then placed in a side-lying position under the stiffness-testing instrument. A full-body vacuum splint (Evac-U-Splint; Hartwell Medical, Carlsbad, California) was secured around the subject's body, and support straps were fastened to the device frame to maintain the test position and mechanical grounding. The test limb was held in 90° of shoulder abduction and 90° of elbow flexion as a rigid body with use of vacuum splints (Hartwell Medical) and was secured to the attachment arm. The shoulder was moved through a range of motion, safety limits were set 10° within the calculated range, and then the shoulder was moved to the test positions at 0° of external rotation and 90% of maximum external rotation. Three repetitions were performed, and the average value was used to determine 50% maximum internal rotation torque for testing.
Stiffness was assessed under four conditions involving two joint positions and two levels of internal rotation torque. The subject was instructed either to relax (resulting in 0% internal rotation torque [the passive condition]) or to maintain an isometric internal rotation torque against the attachment arm (resulting in 50% maximum internal rotation torque [the active condition]). During the passive, or relaxed, tests, an oscilloscope was used to confirm the absence of muscle activity and torque. During the active trials, torque output was displayed and was used as visual feedback to assist the subject in holding 50% maximum internal rotation torque. A perturbation (lasting approximately 60 msec) was randomly applied to quickly displace the subject's arm 5° in an external rotation direction. The movement profile of position and velocity were kept constant to minimize variability and velocity-related damping effects11,12. The subject was instructed not to intervene during the movement. Resistance to the perturbation was detected with the torque-reaction sensor and was used to calculate stiffness as the change in torque (N-m) divided by the change in angular displacement (deg). The test was repeated five times at each joint position and internal rotation torque level with a thirty-second rest between repetitions and a two-minute rest between conditions. All test conditions were presented with use of a Latin squares table for randomization.
Data Analysis
Data were analyzed with use of descriptive and inferential statistics. The stiffness-dependent variable was maximum stiffness (N-m/deg), analyzed with a 2 × 2 (joint position × internal rotation torque level) analysis of variance with repeated measures. The maximum internal rotation torque values (N-m) were analyzed with a paired samples t test. This was performed to determine whether differences in torque generation existed between the two test positions.
The electromyography-dependent variables were the preparatory and reactive values for muscle activity. These variables were analyzed with 2 × 2 × 5 (joint position × internal rotation torque level × muscle) analyses of variance with repeated measures. Independent or paired t tests with Bonferroni correction were used for post hoc analyses.
Laxity scores were converted to nominal data with use of the following operational definitions from previous research. Normal glenohumeral joint laxity was defined as a grade of 1, and abnormal glenohumeral laxity was defined as a grade of 2 or 349. Normal generalized joint laxity was defined as a score of 0 to 3, and abnormal joint laxity was defined as a score of 4 to 949,50. A Pearson product-moment correlation was used to analyze the relationship between generalized joint laxity and glenohumeral joint laxity scores. Separate Pearson product-moment correlations were used to identify whether stiffness and laxity scores (anterior, posterior, inferior, and generalized) were related. The level of significance was set at p = 0.05 for all analyses.
Shoulder Stiffness
When the muscles were contracted to 50% of maximum internal rotation torque, shoulder stiffness was 77% (1.0 - [0.19/0.81 Nm/deg] = 77%) greater than the passive (0°) contraction level, regardless of the joint position (p < 0.001, observed power = 1.00) (Fig. 2). There were no significant differences in stiffness between the joint positions (p = 0.73, observed power = 0.06) or in terms of the contraction level × joint position interaction (p = 0.07, observed power = 0.44) (Fig. 2). There was no significant difference in maximum internal rotation torque between the 0° of external rotation test position and the 90% of maximum external rotation test position (258.6 ± 128.5 compared with 251.5 ± 144.2 Nm; p = 0.052).
Muscle Activity
Two subjects (one male and one female) had poor transmission of electromyography signals; therefore, their data were not used in the electromyography analyses. All analyses of electromyography data represented thirty-eight subjects.
During the preparatory period, there were significant differences in muscle activation area (%·s) between the contraction levels (area of muscle activity in the passive condition = 0.25 ± 0.30%·s, area of muscle activity in the active condition = 0.75 ± 0.95%·s; p < 0.001) and between the muscles tested (p < 0.001). There was significantly more activity of the subscapularis muscle (23% to 37%; [0.27/1.17 to 0.43/1.17]) as compared with the other muscles (p < 0.05), regardless of torque level. Also, a significant two-way interaction effect was found for internal rotation torque level × muscle (p < 0.001) and for joint position × muscle (p < 0.001). On post hoc analysis, the activity of the subscapularis was found to have increased significantly more than that of the other muscles during the active condition (Fig. 3). Last, subscapularis preparatory activity was 36% (1.0 - [0.91/1.43]) higher in the 0° external rotation position than in the 90% of maximum external rotation joint position (p < 0.01) (Fig. 4). No other results were significant (p > 0.05).
During the reactive period, there were significant differences in muscle activation area (%·s) between the contraction levels (area of muscle activity in the passive condition = 1.8 ± 1.1%·s, area of muscle activity in the active condition = 4.6 ± 3.7%·s; p < 0.001) and between the muscles tested (p < 0.001). Post hoc analyses showed significantly less (85% to 93%; [100 - (3.13/0.21) to 100 - (1.39/0.21)]) anterior deltoid activity than other muscle activity following the perturbation. In the active condition, the subscapularis had significantly greater (61% to 86% [1.0 - (4.32/10.94) to 1.0 - (1.55/10.94)]) activity than the other muscles (p < 0.05) (Fig. 3). All of the muscles, except for the anterior deltoid, produced significantly more activity during the active condition. No other results were significant (p > 0.05).
Laxity
Generalized joint laxity (as indicated by a score of =4) was present in 20% of the subjects tested. Glenohumeral joint laxity (as indicated by a grade of =2) was present in the anterior direction in 13% of the subjects, the posterior direction in 15%, and the inferior direction in 15%. Pearson product-moment correlations revealed no relationship between generalized joint laxity and glenohumeral joint laxity (r = 0.03, n = 40, p = 0.005, two-tailed). There was also not a significant correlation between stiffness and laxity (r = -0.12 to 0.29, p = 0.08 to 0.48).
To our knowledge, this is the first study to simultaneously measure in vivo shoulder stiffness and muscle activation in functional positions. Overall, the results indicated that muscle contractions can greatly increase shoulder joint stiffness, regardless of whether the shoulder is positioned at 0° or 90% of maximum external rotation. The subscapularis was activated significantly more than the other muscles; however, its recruitment in the apprehension position (90% of maximum external rotation) was significantly less than that in the neutral position (0° of external rotation). These results suggest that dynamic and static shoulder structures interact to maintain relatively constant levels of joint stiffness and stability, independent of the joint positions tested. Generalized joint laxity and glenohumeral laxity do not appear to correlate with shoulder stiffness in a healthy population.
In the present study, both passive and active stiffness measures were performed. Passive stiffness refers to the resistance against movement provided by static restraints (e.g., ligaments, joint capsule, fascia) and resting muscle tone (i.e., passive muscle tension)8. Active stiffness represents a more functional measure of joint stability, including both the resistance of static structures that are stretched as well as muscle tension that resists changes in joint position8,9. Joint stiffness has been shown to increase as much as tenfold with muscle contractions9,10,51. Zhang et al. reported a sixfold increase in shoulder stiffness with 35 Nm of contraction (40% of maximum contraction)23. Our study confirmed a fourfold increase with a 50% maximum internal rotation torque, demonstrating that large increases in stiffness and stability can be achieved at relatively moderate muscle-activation levels.
We found that stiffness was the same at two different joint positions (0° of external rotation and 90% of maximum external rotation) for both active and passive conditions. With respect to stability, contributions from the joint capsule and ligaments may have a greater and more important role in joint stiffness as they become taut in maximal external rotation. Given the overall characteristics and increased reliance on static structures in external rotation, one may be more vulnerable to anterior instability when the shoulder is in the apprehension position6,52,53. In the middle of the range of motion, the joint capsule and ligaments are relatively loose, emphasizing the importance of muscles (e.g., rotator cuff and scapular stabilizers) for joint stability6,17,18,52. Our results suggest that moderate levels of torque production and stiffness at the shoulder remain relatively constant, although the musculoskeletal structures contributing to anterior-posterior stability may vary18. These findings agree with theories suggesting that consistent levels of stiffness are more desirable because the mechanical behavior of the joint is more predictable and muscle-recruitment strategies can be optimized9,54.
Examination of the muscle-activation strategies is imperative to fully appreciate glenohumeral joint stiffness and stability because joint load and position influence muscle-recruitment levels17,18,55. The results of the present study support a strategy that selectively modifies subscapularis preactivation in anticipation of external rotation loads to optimize energy absorption and to maintain joint stability56-58. In the preparatory and reactive phases, the subscapularis was activated significantly more than the anterior deltoid, supraspinatus, infraspinatus, and teres minor. The subscapularis appears to be responsive to external perturbation and may serve an important role in regulating shoulder stability in external rotation. However, the time-frame of reactive muscle activity of the subscapularis for providing stability in relation to an event of anterior instability remains unknown59,60. Reliance on the subscapularis may also be position-dependent. Turkel et al.17 found through cadaver testing that the subscapularis shifted superior to the humeral head with increased glenohumeral external rotation in 90° of abduction and that the anterior band of the inferior glenohumeral ligament became the primary anterior stabilizer of the humeral head in this injury-prone position17. Theoretically, the glenohumeral joint could be more vulnerable to injury as the static structures contribute more in maintaining stability and the force absorbing capabilities of muscle are minimized18. Our observations also suggest a shift away from the subscapularis as a primary dynamic stabilizer when the shoulder is in the apprehension joint position (90% of maximum external rotation). Interestingly, the level of joint stiffness and stability in the apprehension position was the same as in a neutral position, although the muscle-recruitment strategy had changed. This may be explained by an increase in passive muscle tension, compensatory stiffening of the static structures, or synergistic contribution from other muscles not tested in the present study, such as the pectoralis major and the latissimus dorsi.
Clinically, it is important to understand how both excessive generalized joint laxity and glenohumeral joint laxity may be implicated in the risk of instability and injury19,20,44,56,57. We found no significant relationships with stiffness regulation or muscle-activation strategies, which are in part consistent with the findings of Lintner et al.15. Theoretically, these values should be related, on the basis of the premise that laxity in the static restraints should affect passive joint stiffness and the proprioceptive feedback necessary for coordinated muscle activation, thus altering active stiffness regulation as well57,59,61. However, our physically active subjects were not injured or involved in overhead sports, so there was less opportunity to observe unstable shoulders or those with "acquired" glenohumeral laxity. Our contrasting results may provide normative data and initiate further study to determine the clinical relevance of this theory in different populations. Also, the glenohumeral laxity measures were based on manual tests, performed by a single examiner, that are subjective and that may introduce bias; however, they remain the best clinical method to assess shoulder laxity.
The present study had limitations that should be considered. First, subscapularis muscle activity was measured at a single site. Some researchers suggest differentiating the upper from the lower subscapularis fibers because of their separate innervations62,63. Second, the number of subjects who displayed generalized joint and glenohumeral joint laxity was limited and may have affected the ability to generalize our correlation results. Last, our data reflect those of healthy individuals, and future research should include different populations to maximize external validity.
In conclusion, our results demonstrate that joint stability is regulated with use of both dynamic and static tissues to maintain anterior-posterior joint stiffness in the shoulder against an external rotation load, independent of the joint positions tested in this study. Clinically, healthy populations do not have less stiffness in the apprehension position (90% of maximum external rotation). Dependence on preactivation of internal rotators, specifically, the subscapularis, has an important function for increasing shoulder stiffness, but less preparatory subscapularis activation between neutral and 90% of maximum external rotation positions suggests a transition to other musculoskeletal tissues for stability at the end-range of motion. Last, few incidences of "abnormal" laxity existed in our subjects, and increased generalized joint laxity and glenohumeral joint laxity were not related to shoulder stiffness regulation or muscle-activation strategies in our study population. 