Electromyographic activity of selected shoulder muscles in commonly used therapeutic exercises.

Key Words: Electromyography, Exercise, general, Muscle performance, upper extremity, Shoulder, Upper extremity, shoulder.

Therapeutic exercises are used by physical therapists as an integral component of the rehabilitation of rotator cuff injuries. Exercise protocols are established on the basis of what is known of the functional anatomy and biomechanics of the shoulder complex, previous electromyographic (EMG) studies, and clinical experience. These protocols are primarily directed toward (1) increasing tension-generating capability, flexibility, and endurance in isolated muscles that demonstrate deficiency and (2) restoring the normal kinematics of glenohumeral (GH) and scapulothoracic motion.

To produce normal motion, muscles controlling the shoulder complex must function together in a well-coordinated and efficient manner. Only recently have investigators begun to analyze the patterns of muscle activation in normal, uninjured shoulders during the performance of therapeutic exercises that are routinely used in the clinic.[1,2] The goal of these studies has been to determine which exercises are most useful in activating specific muscles of the shoulder joint complex. Limited information is available evaluating whether patterns of EMG activity are similar to normal patterns when shoulder pathology exists.

In 1982, Jobe and Moynes[3] suggested that supraspinatus muscle activity could be isolated from the activity of other rotator cuff muscles during elevation of the arm in the scapular plane with full medial (internal) rotation of the GH joint, a maneuver commonly referred to as the "empty-can" (EC) exercise. A more recent study by Townsend et al[2] showed similar results. They reported peak activity in the EC exercise to be 74% of that elicited during a maximal voluntary isometric contraction (MVIC), in which subjects performed the exercise in a seated position while holding a light weight. A valuable comparison made in this study was the evaluation of arm elevation in both the frontal and scapular planes. Performing elevation in the frontal plane was not found to significantly activate the supraspinatus muscle according to the criteria used by these authors. These results are in contrast to those reported by Blackburn et al,[4] whose subjects produced the greatest supraspinatus muscle activity when positioned prone while lifting the arm into horizontal abduction at 100 degrees and maintaining extreme lateral (external) rotation. Making direct comparisons between studies is complicated, however, by methodological differences in the quantification of EMG data and the amount of resistance used in each of the exercises.

Infraspinatus and teres minor muscles have been shown to parallel each other in both pattern and extent of activity in many of the movements that have been tested.[2-5] jobe and Moynes[3] found this to be true in sidelying lateral rotation (SIR) with the arm held close to the side and the elbow flexed 90 degrees. Townsend and coworkers[2] supported this finding, reporting that the SIR exercise elicited the most activity in the teres minor muscle and was the second most effective exercise in activating the infraspinatus muscle. Both muscles were active over approximately 50% of the duration of the exercise and reached levels of activation near 80% to 85% of MVIC. Other investigators4 report success in selectively activating the infraspinatus and teres minor muscles in prone lateral rotation (PLR) with the upper arm supported and the elbow flexed to 90 degrees. Similar findings have been reported for other functional and sport-specific movements that require GH abduction with full lateral rotation.[3,6,7]

One of the few studies to include a group of subjects with an identifiable shoulder pathology suggested that neuromuscular imbalance in activating shoulder muscles during the execution of a skilled motor behavior may be an important etiological factor to be considered in evaluating shoulder dysfunction.[6] No studies have been reported that determined whether such an imbalance is detectable in a group of patients involved in a rehabilitation program. The purpose of this study, therefore, was to evaluate and compare muscle activity in the supraspinatus, infraspinatus, teres minor, and lower trapezius musculature in subjects with and without shoulder pathology who performed commonly used therapeutic exercises. This information may be useful to therapists in prescribing the most efficient and effective therapeutic exercise programs designed to ameliorate dysfunction.

Method

Subjects

Subjects for this study were divided into two groups. Twenty subjects (9 men, 11 women) with no history of shoulder pathology comprised one group. A second group consisted of 20 individuals (14 men, 6 women) who reported recurrent unilateral shoulder pain of at least 6 months' duration that limited their occupational or recreational activities. Subjects in this group either were former or current patients treated for shoulder pain at a local outpatient clinic or were recruited from the community. Both groups of subjects ranged in age from 18 to 40 years ([X [bar] = 28, SD = 5.8).

A thorough examination of the shoulder complex, including routine clinical tests of strength, stability, posture, range of motion (ROM), and so forth, was performed on each subject. Criteria for subject selection in the group with shoulder pain included (1) unilateral shoulder pain that persisted for not longer than 24 to 36 hours after strenuous upper-extremity activity and (2) unilateral weakness in lateral rotation and abduction in the scapular plane, with or without clinical signs of GH joint instability. These criteria are consistent with a stage 11 impingement syndrome involving fibrosis and tendinitis of the rotator cuff, as described by Neer.[8]

Weakness was assessed using manual muscle testing (MMT) procedures as outlined by Yocum.[9] Testing of the lateral rotators was done with both arms adducted at the side and the elbows flexed to 90 degrees. Abduction was tested with both arms elevated to 90 degrees in the plane of the scapula while maintaining maximal medial rotation. Unilateral weakness was defined as an inability to maintain the affected arm in the test position against maximum resistance applied at the wrist. All subjects were able to accept moderate resistance on the affected side in each test position, equivalent to an MMT grade of Good according to the grading system of Daniels and Worthingham.[10]

Glenohumeral joint stability was assessed in the anterior, posterior, and inferior directions using the "load and shift maneuver" described by Silliman and Hawkins.[11] These tests involved the application of a directional force to the humeral head with the arm abducted to 60 degrees. With the subject fully relaxed, the examiner palpated the apex of the humeral head as the force was applied and observed the relative amount of passive translation that occurred between the tip of the acromion and the head of the humerus. Using these maneuvers, instability was defined as relative translation of the humeral head 1 to 2 cm from the resting position of the joint before the force was applied.

Among the other factors evaluated were posture, impingement, crepitation, and ROM. Subjects were observed from the front, side, and back for gross postural abnormalities that might relate to shoulder dysfunction. As indicated in Table 1, the most common finding was upper-extremity medial rotation in standing. The impingement tests used were those described by Neer and Welsh[12] and Hawkins and Kennedy.[13] Table 1 presents the positive signs most frequently found in the clinical examinations of subjects with shoulder pathology. None of these patients had signs of acute inflammation or reported a history of subluxation or dislocation. We excluded from participation subjects who had (1) received a cortisone injection in the 2 years prior to testing or (2) undergone any corrective surgical procedures for their shoulder problem. All subjects provided their informed consent prior to participation in the study.

Table 1. Most Frequent Positive
Clinical Signs Among Subjects With
Shoulder Pathology (n = 20)
                                Percentage of
Clinical Sign                   Occurrence
Observation of upper-
  extremity medial
  rotation in standing
  posture                       35
Anterior glenohumeral
  joint instability             40
impingement                     25
Crepitation                     50

Instrumentation

The EMG data were acquired using standard indwelling, fine-wire electrodes.[14] Each electrode consisted of two 50-[mu]m wires(*) that were inserted into a 27-gauge hypodermic needle with 1 to 2 mm of insulation burned from the end. The wires were bent to allow 3 mm of wire to extend back over the beveled tip of the needle. All electrodes were gas sterilized prior to use. The EMG signals were led to a high-impedance (15 m[omega] at 100 Hz) differential amplifier (Model GCS-67([dagger])) following on-site preamplification. The combined preamplifier and main amplifier system permitted a gain of 500 to 10,000 with a bandwidth of 40 Hz to 4 kHz. The common mode rejection ratio was 87 dB at 60 Hz.

A custom-made electrogoniometer was designed using a precision potentiometer([double dagger]) (10 k[omega]) with a lightweight movable arm that was mounted on an adjustable boom. The movable arm was secured to the subject's arm to determine the ROM pertinent to each exercise. The signal from the electrogoniometer was cabled to a DC amplifier (Model ELG-67[dagger]) with an input voltage range of [+ or -]1 V. The electrogoniometer was calibrated to allow conversion of the voltage output to the equivalent degrees of motion. The electrogoniometer signal and EMG data were continuously displayed on an oscilloscopes([section]) to monitor signal quality and recorded on FM tape at a speed of 38.1 cm (15 in) per second. The frequency response of the FM recorded([parallel]) was DC to 5,000 Hz. To assist in confirming proper electrode function, EMG data were input into a Grass audio monitor.(#)

Procedure

Prior to data collection, we determined the weight to be used in each exercise on the basis of hand-held dynamometric measurements. An investigator placed the dynamometer(**) just proximal to the ulnar styloid process when the subject's arm was placed at the end range of each exercise position. The subject was instructed to maintain that position against the examiner's resistance. When the subject was no longer able to overcome the resistance without substitution of other muscles, the value displayed on the dial of the dynamometer was recorded in kilograms of force. Substitution was defined as any deviation in position of any part of the shoulder complex or trunk from the initial isometric position. To standardize the amount of weight used, 25% of the average of three dynamometric measurements in each of the test positions was calculated for each subject. Based on a pilot study, this load was chosen to avoid producing fatigue or pain, particularly in subjects with shoulder pathology, and to minimize the need for substitution. Values obtained in kilograms were converted to pounds and subsequently rounded to the nearest pound.

Following an explanation of the experimental procedure, subjects were positioned prone on a plinth and the skin overlying the muscles to be tested was cleansed with alcohol. The electrodes were inserted into the supraspinatus, infraspinatus, teres minor, and lower trapezius muscles according to the locations described by Delagi and Peroto.[15] After withdrawal of the needle, the free ends of the wires were bared of insulation with the flame of a match and connected to the preamplifier assembly. Each preamplifier was placed approximately 2.5 cm (1 in) from the electrode insertion site, avoiding bony prominences. The reference electrode was placed on the wrist contralateral to the side tested. Prior to testing, we verified electrode placement and adjusted amplifier gain settings by monitoring the EMG signals on an oscilloscope while providing manual resistance to each of the muscles under consideration.

The exercises chosen, primarily on the basis of their clinical use, were prone lateral rotation (PLR), sidelying lateral rotation (SLR), and the "empty-can" (EC) exercise. Exercises were performed in a random order, with subjects completing seven repetitions of each exercise. The PLR exercise required the subject to lie prone with the arm abducted 90 degrees and the elbow at 90 degrees of flexion. The subject moved from neutral to a position of full lateral rotation of the GH joint. In the SLR exercise, the subject was positioned sidelying with the humerus aligned parallel to the trunk and the elbow flexed 90 degrees. From a starting position with the forearm resting on the abdomen, the subject moved to a position of full lateral rotation. The EC exercise was performed with the subject standing. The exercise required the subject to elevate the arm in the scapular plane while maintaining full medial rotation. Subjects were instructed not to elevate their arms beyond 90 degrees.

The timing of each exercise was guided by a metronome([double dagger]) set at 44 beats per minute, which corresponded to approximately 66[degrees]/s, assuming that the arm moved through a 90-degree arc of motion. Subjects practiced each exercise without the weight until performance met velocity requirements. That is, if subjects were completing each repetition of the exercise in reasonably good time with the beat of the metronome, their exercise velocity was judged to be appropriate. The number of practice trials ranged from three to five. Subjects were given verbal commands to begin and end each exercise trial. If EMG or electrogoniometer artifact was detected during a trial, attempts were made to improve signal quality and the trial was repeated. Typically, the aberrant signals were easily identified as randomly occurring, large-amplitude, low-frequency waveforms presumably resulting from movement artifact. If the problem could not be corrected, the data for that muscle were not included in the analyses. In only 5 of the 40 subjects tested, data were not obtained from one of the four muscles of interest in one of the exercises.

Data Reduction

Electromyographic and electrogoniometric signals were transferred from FM tape to a microcomputer([double dagger]) at a sampling rate of 1,000 Hz per channel through a 12-bit analog-to-digital converter([sub-section]) for off-line analysis. DATAPAC II software([double parallel]) was used to full-wave rectify and smooth the EMG data using a 50-millisecond time constant. The same time constant was applied to smooth the electrogoniometric data without prior rectification.

Following data transfer, the processed signals were displayed on a computer monitor. Five phases of exercises using concentric contractions were chosen based on signal quality and consistency of performance, as indicated by total ROM and velocity. The events were selected at the beginning and end of the ROM, requiring concentric muscle contractions on the basis of change in direction of the electrogoniometer trace. Events were marked on the basis of observation, as illustrated in Figure 1, with the phases of concentric contraction bounded by pairs of vertical lines. Five events were chosen in order to obtain an ensemble average of the EMG data from each individual.

The ensemble averages were then divided into three equal time intervals. This was done to better characterize the pattern of response of each muscle to the changing demands for muscle activity during the execution of these exercises. In our view, if between-group differences did exist, those differences would be most apparent at those times when the greatest demands were placed on the muscle. Such differences might otherwise be obscured if only data over the entire period of the exercise were analyzed.

From the averaged signal, mean values for the entire phase of concentric contraction and for each of the three intervals within this phase were calculated and recorded. Each subject's EMG values for each muscle were subsequently normalized as a percentage of the highest mean value in any interval among any of the three exercises performed. For example, if the highest mean EMG activity in the supraspinatus muscle occurred in the third exercise interval of the EC exercise, then the normalized EMG value for that interval would be 100%. Electromyographic values over the entire phase of concentric contraction and in each of the other intervals of all three exercises would be expressed as a proportion of the activity in the third exercise interval.

Because of between-subject differences in the pattern of muscle usage in complex motor activities, the exercise interval in which the highest mean EMG activity occurred in each muscle was anticipated to vary among subjects. An indication of this variability is reflected in the distribution of exercise intervals used to normalize EMG data for each muscle. Table 2 summarizes the locations of the intervals most frequently found to elicit the highest EMG activity for each muscle. As is shown, the highest EMG values generally occurred in one of only three exercise intervals for each muscle. Most commonly, this occurred in the third interval of either the EC or PLR exercise, as would be predicted because the influence of gravity was at its maximum in these positions.

[TABULAR DATA 2 OMITTED]

Data Analysis

A two-way (group X ROM) repeated-measures analysis of variance (ANOVA) was used to assess significant differences in the total ROM between subject groups in each of the three exercises. To evaluate between-group differences in EMG among the three exercises, a two-way (group X exercise) repeated-measures ANOVA procedure was performed separately for each muscle. For these analyses, the mean normalized EMG values over the entire averaged ROM were compared across exercises for each muscle.

The same repeated-measures design (group X interval) was used for each exercise to detect differences in the level of muscle activity across exercise intervals for each muscle. The mean normalized EMG within each of these intervals was used to determine differences in muscle activity between groups and among intervals. Because across-muscle comparisons were not appropriate, analyses for each muscle and each exercise were conducted separately, thus yielding 12 additional ANOVAS (four muscles X three exercises). Specific differences were determined from a post hoc analysis using a Bonferroni adjusted t test for multiple pair-wise comparisons while maintaining an alpha level of .05 for significance in each of the analyses.

Results

Statistical analysis of the ROM data provided by the electrogoniometer revealed no significant differences in ROM between groups or across exercises. Table 3 presents the descriptive statistics for this portion of the analysis.

[TABULAR DATA 3 OMITTED]

In analyzing the mean EMG activity over the entire phase of concentric contraction of each exercise, no significant group difference was revealed in any of the muscles tested. A significant effect due to exercise, however, was found in all four muscles. These data are summarized in Figure 2, showing which exercises were most effective in producing activity in each muscle. To evaluate how the muscle was activated within each exercise, mean normalized EMG values for the three intervals of an exercise were compared in each of the muscles tested. No group difference was found in any of the muscles evaluated in the EC exercise. The data, therefore, were averaged across all subjects. The ANOVAs revealed a significant main effect for exercise interval for all muscles. In the EC exercise, both the supraspinatus and lower trapezius muscles showed progressively increasing activity throughout the ROM (Fig. 3). Infraspinatus and teres minor muscles were only minimally active (< 10%) in this exercise.

Similarly, the analyses for the SLR exercise demonstrated effects due only to the exercise interval for each of the muscles tested, without a significant group effect. Relatively moderate amounts (20%-60%) of EMG activity were elicited in each of the four muscles, with the infraspinatus and teres minor muscles being most active in the second and third exercise intervals Fig. 4).

Table 4 summarizes the results of the pair-wise comparisons between intervals for each muscle and each exercise. Only the probability values for those interval comparisons for which a significant difference was found are reported.

[TABULAR DATA 4 OMITTED]

Shown graphically in Figure 5 are the results of the group X interval analyses of the PLR exercise for each muscle. A progressively increasing pattern of activity was observed throughout the ROM in each of the four muscles. Table 4 shows that in the PLR exercise, EMG activity was different in each of the interval comparisons for all of the muscles under investigation. Although the presence of pathology did not influence muscle activity in either the EC or SLR exercise, the PLR exercise showed a group effect in both the supraspinatus and infraspinatus muscles (Figs. 5A and 5B). The ANOVA and subsequent follow-up analysis for the supraspinatus muscle revealed a group X interval interaction in which the EMG activity in intervals 2 and 3 was higher in the subjects without shoulder pathology than in the subjects with shoulder pathology. In the analysis for the infraspinatus muscle, a significant group main effect indicated that the mean normalized EMG elicited during the PLR exercise was consistently higher in the subjects with shoulder pathology than in the subjects without shoulder pathology throughout the entire ROM. No group effect was demonstrated for either the teres minor or lower trapezius muscles in this exercise (Figs. 5C and 5D).

Discussion

The ROM data from the electrogoniometer were assessed in order to establish that EMG values from the ensemble averages were analyzed over comparable portions of the ROM in each exercise. As shown in Table 3, the greatest variability was observed in the SLR exercise. This result was anticipated because the starting and ending positions were least well-defined for this exercise, depending heavily on subject comfort and flexibility.

A major factor contributing to variability was that five subjects without symptoms and four subjects with shoulder pathology had ROM values between 100 and 125 degrees for the SLR exercise. Compared with other members of their respective groups, these subjects began the exercise in a position of relatively more medial rotation and ended in one of greater relative lateral rotation. Although this resulted in a larger arc of motion for these individuals, the degree of overlap with intervals used for other subjects was not considered significant and we believe this should not have affected the overall interpretation of the EMG results. Similar circumstances explain the variability in the other two exercises.

The choice of exercises included in this investigation was based in part on previous work demonstrating the relative specificity of each exercise in activating the muscles of interest. Because of the limited number of exercises, no inferences should be made regarding the optimal exercise for a given muscle. The relative effectiveness of each of the exercises in eliciting muscle activity, however, is shown in the results of the group X exercise analyses, which generally confirm earlier reports found in the literature (Fig. 2).

As expected, the EC exercise was clearly the most effective in activating the supraspinatus muscle. Of the 16 exercises evaluated by Townsend et al,[2] the EC exercise was second only to the military press exercise in eliciting supraspinatus muscle activity, thereby supporting our result. In the lower trapezius muscle, the PLR exercise elicited the most activity, whereas the EC exercise was least effective for this muscle. This trend reflects the role of the lower trapezius muscle as a stabilizer of the scapula and agrees favorably with those researchers who argue that this muscle participates minimally in GH abduction below 90 degrees.[1,16-18] Although Moseley et al[1] reported that SLR did not qualify as an exercise that significantly challenged the lower trapezius muscle, the amount of resistance used was not standardized across subjects or across exercises within a subject and may account for the result that differs from our own. No previous report of lower trapezius muscle activity during PLR could be found.

The PLR and SLR exercises proved equally effective in eliciting activity from the infraspinatus and teres minor muscles, reflecting their functional roles as prime movers in lateral rotation. These same exercises were evaluated by Blackburn et al,[4] who reported significantly greater activity in infraspinatus and teres minor muscles in the PLR exercise than in the SLR exercise. Their result, however, may have been due to the method used to quantify EMG activity, as comparisons of muscle activity were made between exercises only at a time when the arm was perpendicular to gravity. Thus, EMG activity in the final position of the PLR exercise was compared with that in the middle of the ROM of the SLR exercise, a method that in our opinion severely limits the interpretation of those findings.

The results of the group X interval analyses show that the general pattern of muscles activity within each of the three exercises was similar for subjects with and without shoulder pathology. For each muscle, a characteristic pattern of activity was demonstrated in each of the exercises studied. Only one of the three exercises (ie, PLR) showed significant between-group differences. These differences were related to the extent to which the supraspinatus and infraspinatus muscles were activated in normal as compared with pathologic shoulders. Subjects with shoulder pathology were found to have significantly greater activity in the infraspinatus muscle and less activity in the supraspinatus muscle as the GH joint was laterally rotated in a prone position (Figs. 5A and 5B). In our view, these findings suggest an imbalance in the activation of the posterior rotator cuff muscles during specific shoulder movements. Our analysis, however, was of EMG values, rather than ratios, and our data therefore did not directly examine the issue of imbalance (relative values). Whether an imbalance in muscle activity is part of the primary pathology contributing to shoulder dysfunction or is secondary to some other pathology cannot be discerned from the results of this study.

Imbalances have been identified in pathologic shoulder motion in other studies, either in the form of torque ratios or EMG activity. Using isokinetic strength testing, Warner and colleagues[19] showed that medial/lateral rotation torque ratios were higher in the shoulders of patients classified as either having GH joint instability or impingement problems. In addition, an association between lateral rotator weakness, anterior GH joint instability, and impingement syndrome was found. A similar association was characteristic of the subjects with shoulder pathology in our study. All of the subjects with shoulder pathology demonstrated asymmetrical weakness in MMT of the lateral rotators, and four of the eight subjects who exhibited positive clinical signs of anterior GH joint instability also showed a positive sign for impingement.

Lending further support for our hypothesis, Glousman and co-workers[6] reported an imbalance in muscle activity in a group of athletes with chronic anterior instability of the GH joint in various shoulder muscles throughout the five phases of a throwing motion. Of particular interest was the finding that athletes with GH joint instability demonstrated significantly greater infraspinatus muscle EMG activity during the early cocking phase of throwing, a motion somewhat analogous to the PLR exercise in our study in that the arm is maintained in an abducted position as the GH joint is laterally rotated.

In addition to muscle imbalance, impingement and instability problems have also been associated with improper mechanics of the GH joint. Howell et al[20] demonstrated that the normal posterior translation of the humeral head that accompanies lateral rotation of the GH joint was absent in a group of patients with anterior shoulder instability. Previous authors[21-23] have proposed that the posterior rotator cuff musculature plays a critical role in the dynamic stabilization of the GH joint.

Cain et al[23] have suggested that, in addition to controlling lateral rotation, the infraspinatus and teres minor muscles are capable of reducing the strain in passive structures that constrain the motion of the humeral head in the glenoid fossa by pulling the humeral head posteriorly in the normal shoulder. In the presence of instability, the lateral rotators might be expected to increase activity in an attempt to position the humeral head appropriately in the glenoid fossa, thereby preventing an already unstable joint from producing excessive strain of the ligamentous capsule and other soft tissues surrounding the joint. Such an explanation seems tenable to account for the increase in infraspinatus muscle activity observed in our study, given the proportion of subjects with shoulder pathology who demonstrated anterior GH joint instability.

Contributing further to the loss of stability of the GH joint was the diminished activity found in the supraspinatus muscle in the PLR exercise. In the third interval of PLR, mean normalized supraspinatus muscle EMG activity was 33% lower in subjects with shoulder pathology, This finding may represent a pain-inhibited response to impingement of that muscles' tendon. From an anatomical perspective, the supraspinatus muscle would be more vulnerable to impingement in PLR than in either of the other exercises. In SLR, the arm remained adducted at the side of the body, a position in which lateral rotation would not be expected to compromise the area of the coracoacromial space. Similarly, in the EC exercise, the arm was maintained in maximal medial rotation, thereby placing the insertion of the supraspinatus muscle's tendon out from under the coracoacromial arch and decreasing the risk of impingement on painful tissue. In contrast, the PLR exercise placed the tendon directly under the acromion as the concentric phase of the exercise progressed. Therefore, a pain-inhibited response should be considered a possible factor contributing to the significantly lower EMG activity observed in the second and third intervals of this exercise.

The appearance of an imbalance in muscle activation in only one of the three exercises investigated suggests that further research is needed to determine whether these differences are a common finding among subjects with shoulder pathology. In particular, studies that show similar differences in muscle activity during more functional movements would be useful in determining whether these relationships have meaningful implications in daily activities. Interestingly, an imbalance was found only in the exercise that placed the arm in an abducted, laterally rotated position. This finding would suggest that future studies examining functional movements that present a similar challenge to the stability of the GH joint may be most likely to reveal relationships that resemble those found in these simple exercises.

Conclusions

The results of this study showed that persons with and without rotator cuff pathology use similar patterns of muscle activity to perform common therapeutic exercises. The unique between-group differences found in the relative extent to which the supraspinatus and infraspinatus muscles were activated during the PLR exercise suggest that an imbalance in neuromuscular control may be a factor contributing to shoulder dysfunction in some individuals. Such an imbalance could either be part of the primary pathology leading to dysfunction or a compensatory strategy related to instability or impingement. Because differences were revealed in only one of the three exercises studied, further investigation is needed to determine whether these relationships have meaningful implications in more functional movements. If these relationships prove to be a common finding among subjects with shoulder pathology, future prospective studies that investigate various treatments applied to this population would be useful to assist in developing more specific protocols designed to correct dysfunction.

References

[1] Moseley BJ, Jobe FW, Pink M, et al. EMG analysis of the scapular muscles during a shoulder rehabilitation program. Am J Sports Med. 1992;20:128-134. [2] Townsend H, Jobe FW, Pink M, Med. J. Electromyographic analysis of the glenohumeral muscles during a baseball rehabilitation program. Am J Sports Med. 1991;19:264-272. [3] Jobe FW, Moynes DR. Delineation of diagnostic criteria and a rehabilitation program for rotator cuff injuries. Am J Sports Med. 1982;10: 336-339. [4] Blackburn TA, McLeod WD, White B, Wofford L. EMG analysis of posterior rotator cuff exercises. Athletic Training. 1990;25:40-45. [5] Gowan ID, Jobe FW, Tibone JE, et al. A comparative electromyographic analysis of the shoulder during pitching: professional versus amateur pitchers. Am J Sports Med. 1987;15: 586-590. [6] Glousman R, Jobe FW, Tibone J, et al. Dynamic electromyographic analysis of the throwing shoulder with glenohumeral instability. J Bone Joint Surg [Am]. 1988;70:220-226. [7] Kronberg M, Nemeth G, Brostrom L. Muscle activity and coordination in the normal shoulder. Clin Orthop. 1990;257:76-85. [8] Neer CS. Impingement lesions. Clin Orthop. 1983;173:70-77. [9] Yocum LA. Assessing the shoulder: history, physical examination, differential diagnosis, and special tests used. Clin Sports Med. 1983;2: 281-289. [10] Daniels L, Worthingham C. Muscle Testing: Techniques of Manual Examination. 4th ed. Philadelphia, Pa: WB Saunders Co; 1972. [11] Silliman JF, Hawkins RJ. Current concepts and recent advances in the athlete's shoulder. Clin Sports Med. 1991;10:693-705. [12] Neer CS, Welsh RP. The shoulder in sports. Orthop Clin North Am. 1977;8:583-591. [13] Hawkins RJ, Kennedy JC. Impingement syndrome in athletics. Am J Sports Med. 1980;8: 151-163. [14] Basmajian JV, Stecko G. A new bipolar electrode for electromyography. J Appl Physiol. 1962; 17:849. [15] Delagi EF, Peroto A. Anatomic Guide for the Electromyographer. Springfield, III: Charles C Thomas, Publisher; 1980. [16] Inman VT, Saunders M, Abbot LC. Observations on the function of the shoulder joint. J Bone Joint Surg [Am]. 1944;26:1-30. [17] Weidenbauer MM, Mortenson OA. An electromyographic study of the trapezius muscle. Am J Phys Med. 1957;31:363-371. [18] Bagg S, Forrest W. Electromyographic study of the scapular rotators during arm abducuton in the scapular plane. Am J Phys Med. 1986;65:11-124. [19] Warner JJP, Micheli LJ, Arslanian LE, et al. Patterns of flexibility, laxity, and strength in normal shoulders and shoulders with instability and impingement. Am J Sports Med. 1990; 18:366-375. [20] Howell SM, Galinat BJ, Renzi AJ, Marone PJ. Normal and abnormal mechanics of the glenohumeral joint in the horizontal plane. J Bone Joint Surg [Am]. 1988;68:398-404. [21] Saha AK. Dynamic stability of the glenohumeral Joint. Acta Orthop Scand. 1971;42: 491-505. [22] Perry J. Anatomy and biomechanics of the shoulder in throwing, swimming, gymnastics, and tennis. Clin Sports Med. 1983;2:247-270. [23] Cain PR, Mutschler JA, Fu FH, Lee SK. Anterior stability of the glenohumeral joint: a dynamic model. Am J Sports Med. 1987;15:144-148.

Commentaries

Following are two commentaries on "Electromyographic Activity of Selected Shoulder Muscles in Commonly Used Therapeutic Exercises."

Ballantyne and colleagues collected data to support the argument that patients with glenohumeral joint pathology demonstrate a different pattern of electromyographic (EMG) activity of some shoulder muscles during selected movements as compared with asymptomatic subjects. The authors concluded, therefore, that an imbalance may exist in the pattern of muscle activation during some shoulder movements in patients with shoulder pain. We believe, however, that there are several issues related to the design of this study that need to be considered when interpreting the meaningfulness of the authors' conclusion. Our queries to the authors relate to what appears to be an unclear rationale for subject selection and, most importantly, to the development of a theoretical argument for explaining the meaningfulness of the results.

The method of patient selection appears to be unclear for two reasons. First, the subjects in the patient group were found by the authors to have a variety of impairments, which may have indicated the presence of numerous pathologies. The heterogeneity of the patient sample is an important issue because the authors in their discussion attempt to explain some of their results based on an impairment (glenohumeral joint instability) that they felt existed in only four of their subjects. Given a more homogeneous sample, the authors may have been able to better explain their results. Second, the reliability of measurements used for subject inclusion is suspect. The authors required that subjects in the patient group had unilateral weakness of the abductors and lateral (external) rotators but used a method of muscle testing that has no demonstrated evidence of reliability within the grade range used. There are data and theoretical arguments to support the notion that manual muscle test judgments above the grade of Fair are unreliable.[1,2] We are curious as to why the authors chose not to use an instrument such as a hand-held dynamometer to identify patients with weakness. There are some data to support the reliability of measurements of the shoulder obtained with a hand-held dynamometer.[3,4]

Many of the other examination procedures used for describing the patient group (see Tab. 1 in the article) have no demonstrated evidence of reliability (eg, instability tests, impingement tests). We would argue that classifying patients as having an instability or an impingement syndrome does not appear to be sound, based on the methods used by the authors. other examples of unclear procedures used to select patients can be found in the article. For example, the authors restricted admission of subjects with "signs of acute inflammation" but did not define what they meant by this term.

We have addressed measurement issues and other theoretical issues related to patient selection because we believe that a complete description of the subjects tested is necessary for future research in this area. Given the lack of reliability and the heterogeneity of the sample, we do not believe this study is replicable.

An issue directly related to subject selection is the theoretical construct for the meaningfulness of the question asked in this study. If the purpose of the study was to compare subjects with and without shoulder pathology to determine whether a pathologic pattern of myoelectric activity can be observed, then we are not sure what theoretical or empirical model was being used to explain the results. Why should physical therapists be interested in whether patterns of EMG activity are similar or different when these two groups are compared? Are the authors interested in developing the EMG technique used in this study as a diagnostic tool? We believe the potential impact this report will have on practice rests with the theoretical argument for the usefulness of the question.

The introduction began by discussing the importance of shoulder muscles working in groups to produce "well-coordinated" motion. The authors then discussed those studies that attempted to determine which exercises produce the greatest electrical activity in specific muscles. We are not sure how these two concepts (coordinated activity of muscle groups and optimal EMG activity of a given muscle during an exercise) relate to the authors' question. We hope the authors can clarify for us the theoretical construct that explains why their question is clinically important.

Although there were no formal hypotheses stated in the introduction, the authors implied they had a hypothesis that explained the relationship between the amount of myoelectric activity in the two groups. The authors hypothesized in the "Data Reduction" section of the report that if between-group differences existed, the differences would be most apparent when the greatest demands were placed on the muscle. The authors, however, found that one of the two between-group differences occurred in the supraspinatus muscle during the prone lateral rotation (PLR) exercise, an exercise that did not place the greatest demand on the supraspinatus muscle. The other between-group difference was found to occur in the infraspinatus muscle during the PLR exercise. The patient group had significantly higher EMG activity during this exercise. However, the infraspinatus muscle's EMG activity during the PLR exercise was no different than that found during the sidelying lateral rotation (SLR) exercise in the patient group. There was no difference in infraspinatus muscle EMG activity between the two groups during the SLR exercise. These findings would appear to conflict with the hypothesis implied by the authors.

The authors claimed that the difference between groups in supraspinatus muscle EMG activity during the PLR exercise was due to positional factors. The supraspinatus tendon was said to be more vulnerable to impingement during the PLR exercise and therefore may have demonstrated a "pain-inhibited response." The authors did not define what they meant by the term "pain-inhibited response." In addition, the authors provided no data to support the notion that the position of the limb during the exercise resulted in the differences that were found. The authors did not report whether the subjects reported pain during the exercise sessions.

One of the articles referenced by the authors to argue for the importance of the study was the article by Glousman and colleagues.[5] Ballantyne and colleagues further claimed that their results are consistent with those of Glousman et al.

Glousman and colleagues compared the shoulder muscle EMG activity of two groups of skilled baseball players during the pitching motion. One group consisted of healthy subjects, whereas the other group consisted of subjects with surgically confirmed glenohumeral instability. Glousman and colleagues found significantly greater infraspinatus muscle EMG activity in the patients during the early cocking phase of throwing, a position Ballantyne et al argue is similar to the position attained during the PLR exercise. However, this difference between the groups during the early cocking phase was not reported by Glousman et al to be statistically significant.

It would also appear that the late cocking phase, as defined by Glousman et al, most closely corresponds to the end-range position of the PLR exercise (interval 3) studied by Ballantyne et al. in the late cocking phase, Glousman et al found, unlike Ballantyne et al, that the infraspinatus muscle activity of the asymptomatic subjects exceeded that of the patient group, although not significantly. Also of interest is the fact that Glousman et al found that the supraspinatus muscle activity of the patients during the late cocking phase was significantly greater than that of the healthy subjects. This finding would also appear to contradict the results of Ballantyne et al, who found that supraspinatus muscle activity of the asymptomatic subjects was significantly greater than that of the patients when in a similar position (interval 3). The data of Glousman et al appear to suggest that EMG activity of the supraspinatus and infraspinatus muscles in the two groups is not dependent on the position of the shoulder in the same way as that reported by Ballantyne et al. The apparent paradox between the results of Ballantyne et al and those of Glousman et al may be more presumed than real. The types of movements studied and the methodologies of the two studies differed in many ways and probably provide adequate justification to restrict direct comparison. For example, Glousman et al studied a homogeneous group of patients with surgically confirmed instability, whereas Ballantyne et al studied a patient group that appeared to have a large variety of diagnoses. These two studies, however, provide most of the basis for assessing the usefulness of EMG data in identifying differences between healthy subjects and patients with shoulder pathology during defined movements. At this point, the data would appear to us to be inconclusive.

Ballantyne et al reviewed literature in the introduction that examined which exercises produced the highest EMG activity in specific muscles. We expected to find a stronger argument in the introduction for why differences in EMG activity between healthy subjects and patients during specific exercises might be useful when determining treatment. We are unclear as to how this information would be useful for therapists designing exercise programs for their patients.

Daniel L Riddle, PT Associate Professor Department of Physical Therapy School of Allied Health Professions Medical College of Virginia Virginia Commonwealth University Box 224 MCV Station Richmond, VA 23298

Robert L Lamb, PhD, PT Professor and Chairman Department of Physical Therapy School of Allied Health Professions Medical College of Virginia Virginia Commonwealth University

References

[1] Frese E, Brown M, Norton BJ. Clinical reliability of manual muscle testing: middle trapezius and gluteus medius muscles. Phys Ther. 1987;67:1072-1076. [2] Iddings DM, Smith LK, Spencer WA. Muscle testing, part 2: reliability in clinical use. Phys Ther 1961;41:249-256. [3] Byl NN, Richards S, Asturias J. Intrarater and interrater reliability of strength measurements of the biceps and deltoid using a hand-held dynamometer. J Orthop Sports Phys Ther. 1988; 9:399-405. [4] Sullivan SJ, Chesley A, Hebert G, et al. The validity and reliability of hand-held dynamometry in assessing isometric external rotator performance. J Orthop Sports Phys Ther. 1988; 10:213-217. [5] Glousman R, Jobe FW, Tibone J, et al. Dynamic electromyographic analysis of the throwing shoulder with glenohumeral instability. J Bone Joint Surg [Am]. 1988;70:220-226.

I would like to compliment the authors on an interesting and informative preliminary study, and I encourage them to pursue this topic further.

The authors have identified, through electromyographic (EMG) analysis, a characteristic pattern of muscle activity during three commonly performed rehabilitative exercises. In addition, they have noted in their symptomatic shoulder group a significant alteration in the magnitude of the EMG activity of the supraspinatus and infraspinatus muscles during prone lateral (external) rotation (PLR). The authors have reported in this group greater EMG activity in the infraspinatus muscle and a decrease in EMG activity in the supraspinatus muscle when compared with the control group. They have suggested that this is an imbalance in the activation of the posterior rotator cuff muscles during specific movements when the shoulder exhibits instability or impingement.

There are several points I would like to discuss regarding this study. The authors have equally divided the subjects into two groups, one consisting of subjects with symptomatic shoulders and the other of subjects with asymptomatic shoulders. It is not clear, however, as to exactly what type of shoulder pathology these individuals exhibited. In Table 1, the distribution is such that 8 out of 20 subjects with shoulder pathology exhibited anterior instability, and an additional 5 subjects exhibited a positive impingement sign, but what diagnosis did the remaining subjects in this group exhibit? Also, for the subjects exhibiting impingement, was this a primary or secondary pathology due to instability. I feel it is important to delineate the clinical conditions, such as impingement or instability. The primary pathology should be identified because of the possible alterations in normal glenohumeral joint kinematics and specific muscular imbalances. Such an example exists with patients who have stage II impingement, as described in this report. We have identified that this type of patient exhibits a reverse capsular pattern, whereas medial (internal) rotation is restricted more than elevation, which is limited more than lateral rotation.[1] Several authors[1-3] have noted that posterior capsular tightness can lead to altered glenohumeral kinematics and may cause the humeral head to migrate anterosuperiorly during arm movements above the horizontal plane. This anterosuperior migration of the humeral head can cause impingement on the coracohumeral ligament or the anterolateral surface of the acromion, and thus cause a painful arc.

This may help to explain the decrease in EMG activity of the supraspinatus muscle during PLR in subjects with impingement. The alteration in EMG activity was not noted during the other two movements.

Additionally, 40% of the subjects with symptoms exhibited anterior glenohumeral instability. None had a history of subluxation or traumatic dislocation; therefore, these individuals can be classified as having atraumatic instability and probably exhibited a degree of excessive laxity. Often individuals with excessive laxity of the shoulder who complain of recurrent pain without gross instability are actually silent subluxators.[4] Jobe and colleagues[4,5] state that the most sensitive means for detecting silent subluxations of the glenohumeral joint is the use of an apprehension test followed by a relocation test. In younger populations (ages 18-27 years), this subtle clinical condition can lead to impingement and other types of pathology, all secondary to instability.

I perform anteroposterior ligamentous stability tests at 0, 45, and 90 degrees of abduction to assess specific ligamentous structures of the glenohumeral capsule.[6-8] I also perform an inferior translation test at 0, 45, and 90 degrees of abduction to test the superior glenohumeral ligament and the anterior and posterior bands of the inferior glenohumeral ligament, respectively.[9] I believe these tests are important in determining multidirectional instability, subtle subluxations, and the overall degree of laxity.

The authors have addressed a key point of dynamic shoulder stability: muscular imbalance. Stability of the glenohumeral joint is accomplished through two distinct types of mechanisms: active and passive. The passive mechanisms include the joint's geometric configuration, ligamentous restraints, glenoid labrum, soft tissue barrier, limited joint volume, and the adhesion/cohesion principle of articulating surfaces. The active mechanism of glenohumeral joint stability is the "dynamic stabilizers," which include the muscles surrounding the shoulder complex and the innervation to those muscles. The dynamic stabilizers accomplish dynamic stability through active compression of joint surfaces and dynamic ligament tension. Additionally, the dynamic stabilizers provide stability through an active mechanism referred to as "neuromuscular control." I strongly believe that neuromuscular control is the essential element to functional stability of the glenohumeral joint. The glenohumeral joint is greatly dependent on the dynamic stabilizers for stability due to its inherently inadequate stability. Dynamic glenohumeral joint stability is thought to be accomplished through proprioceptive/kinesthetic awareness and efficient muscular activity.[10] Dynamic stability is believed to be achieved through the efficiency of several muscular force couples about the shoulder.[11] These force couples are the anterior deltoid muscle counter-balanced by the infraspinatus and teres minor muscles and the subscapularis muscle counterbalanced by die infraspinatus and teres minor muscles.[11,12] Dynamic glenohumeral stability is accomplished by these force couples through joint compression,[13,14] dynamic ligament tension,[15] and proficient neuromuscular control.[10]

The investigators of this study have begun exploration of the concept of muscular imbalance during dynamic exercises. I believe the three exercises performed in this study represent a hierarchy of shoulder exercises. The "empty-can" exercise is a low-level exercise, that is, one that requires minimal dynamic stability and minimal activity of the force couples to maintain glenohumeral stability. This could explain the low-level EMG activity of the infraspinatus and teres minor muscles during the empty-can exercise. I would consider the next exercise--sidelying lateral rotation--a low- to moderate-level dynamic stabilizing exercise. In this position (ie, sidelying with an adducted arm), inherent glenohumeral stability is enhanced through gravitational forces, adhesion and cohesion of joint surfaces, ligamentous stability, and other factors. Thus, during these two exercises, minimal muscular activity is necessary to dynamically stabilize the glenohumeral joint through co-contractions. Additionally, during these movements, the prime movers (supraspinatus and lateral rotator muscles, respectfully) can contract with only minimal involvement from other muscles. In contrast, the PLR position is a more challenging position. It compromises the static stabilizers and requires enhanced dynamic muscular stability. This may contribute to the explanation of consistently higher infraspinatus muscle activity in the shoulder pathology group in an attempt to dynamically stabilize the humeral head. I wonder whether the subjects' shoulder was supported by the table, or was the shoulder unsupported and over the side of the table? This subtle difference may contribute to significant differences in EMG activity.

The investigators have done an excellent job and should be complimented on their work. Although the group with pathology was not well defined, I think this study is not only interesting but compatible with studies by other investigators.[16,17] This study also has significant merit. It is one of the first studies to investigate, with EMG, muscular activity during specific exercises among asymptomatic and symptomatic subjects. Additional studies in the area of muscular imbalances are necessary. This study has contributed to the explanation of dynamic shoulder stability. I would strongly recommend further exploration of this topic with well-defined groups, using EMG data of the primary force couples of the glenohumeral joint musculature during various specific exercises and functional movements. I agree with the authors that exercise protocols are directed toward increasing tension-generating capability and endurance and restoring normal kinematics. I believe, however, that an essential goal of our therapeutic exercise should be directed toward enhancing our patients' neuromuscular control and thus improving dynamic functional stability of the glenohumeral joint.

Kevin E Wilk, PT Director of Research and Clinical

Education HEALTHSOUTH Rehabilitation Corp Two Perimeter Park S, Ste 224 W Birmingham, AL 35243

Associate Clinical Director HEALTHSOUTH Sports Medicine and

Rehabilitation Center 1201 11tb Ave S, Ste 100 Birmingham, AL 35205

Director of Rehabilitative Research American Sports Medicine Institute Birmingham, AL

References

[1] Wilk KE, Andrews JR. Rehabilitation following subacromial decompression. Orthopedics. 1993;16:349-358. [2] Matsen FA, Arntz CT. Subacromial impingement. In: Rockwood CA, Matsen FA, eds. The Shoulder Philadelphia, Pa: WB Saunders Co; 1990:623-646. [3] Harryman DT, Sidles JA, Clark JM. Translation of the humeral head on the glenoid with passive glenohumeral motion. J Bone Joint Surg [Am]. 1990;72:1334-1343. [4] Jobe FW, Tibone JE, Jobe CM, Kvitne RS. The shoulder in sports. In: Rockwood CA, Matsen FA, eds. The Shoulder. Philadelphia, Pa; WB Saunders Co; 1990:961-990. [5] Jobe FW, Bradley JP. The diagnosis and nonoperative treatment of shoulder injuries in athletes. Clin Sports Med. 1989;8:419-437. [6] Wilk KE, Andrews JR. Current Concepts in the Evaluation and Treatment of Shoulder Instability. Alexandria, Va: Section on Orthopaedics, American Physical Therapy Association; 1993: chap 5. [7] O'Brien SJ, Neves MC, Arnoczky SP, et al. The anatomy and histology of the inferior glenohumeral complex of the shoulder. Am J Sports Med. 1990;18:449-456. [8] Bowen MK, Warren RF. Ligamentous control of shoulder stability based on selective cutting and static translation experiments. Clin Sports Med 1991;10:757-782. [9] Warner JP, Deng X, Warren RF, Torzilli PA. Static capsuloligamentous restraints to superior-inferior translation of the glenohumeral joint. Am J Sports Med. 1992;20:675-685. [10] Wilk KE, Arrigo CA. Current concepts in the rehabilitation of the athletic shoulder. J Orthop Sports Phys Ther In press. [11] Inman VT, Saunders JR, Abbott JC. Observations on the function of the shoulder joint. J Bone Joint Surg. 1944;26:1-30. [12] Basmajian JV, DeLuca CJ. Muscles Alive 5th ed. Baltimore, Md: Williams & Wilkins; 1985: 270-271. [13] Saha AK. Dynamic stability of the glenohumeral joint. Acta Orthop Scand. 1971;42: 491-505. [14] Howell SM, Galinet BJ. The glenoid labral socket: a constrained articular surface. Clin Orthop. 1989;243:122-126. [15] Cleland J. On the actions of muscles passing over more than one joint. J Anat Physiol. 1866;1:85-93. [16] Glousman R, Jobe FW, Tibone J, et al. Dynamic electromoyographic analysis of the throwing shoulder with glenohumeral instability. J Bone Joint Surg [Am]. 1988;70:220-226. [17] Warner JP, Micheli LJ, Arslanian LE, et al. Patterns of flexibility, laxity, and strength in normal shoulders and shoulders with instability and impingement. Am J Sports Med. 1990; 18:366-375.

Author Response

We greatly appreciate the thoughtful commentaries of Mr Wilk, Mr Riddle, and Dr Lamb. The issues that are raised demonstrate the need for further work in this area and provide valuable guidelines for future studies exploring this topic. Although considerable attention has been paid to neuromuscular control related to pathologies such as patellofemoral pain,[1] the concept has been largely ignored in the literature related to the shoulder complex. In this preliminary study, we have attempted to identify what we believe is an important factor to consider in the etiology of shoulder dysfunction.

An area of concern posed in both commentaries is with regard to our method for subject selection in the group with shoulder pathology. The medical diagnoses for subjects in this group varied widely and included nonspecific "shoulder pain," chronic tendinitis, chronic bursitis, and impingement syndrome. Our first criterion for selecting subjects into this group, as stated in our article, was unilateral weakness in lateral rotation and abduction in the scapular plane that was detectable by manual muscle test. When testing both shoulders simultaneously, as is routinely done in the clinic, these weaknesses were easily discerned. In our view, further quantification of these deficits would have yielded little additional useful information related to the purpose of our study.

The second criterion for selection was that the affected shoulder functioned adequately for light activity but was recurrently symptomatic after vigorous use, particularly in overhead activities. Such subjective complaints are consistent with those of a stage II impingement lesion with chronic tendinitis and fibrosis of the rotator cuff.[2] As pointed out in each of the commentaries, only 25% of our subjects displayed clinical signs of impingement. These findings may relate to the relative acuteness of the pathology at the time our evaluation was conducted. Had our subjects been evaluated during an exacerbation of their symptoms, at a time when the lesion was more irritable, we strongly suspect that a greater proportion would have had a positive impingement test result. The presence or absence of physical signs of impingement or instability, however, were findings of our own clinical examination for each subject after having met our criteria for selection. Though the reliability of the examination procedures used in our study may not be proven, this has not precluded their routine use in the clinic as tools for decision making. Studies showing the reliability of procedures for shoulder examination such as those suggested here by Wilk and elsewhere by others[3,4] are needed to provide better standards on which to base those decisions, and we agree that more extensive studies with well-defined subject groups are needed to clarify the relationship between muscular imbalance and specific shoulder pathologies.

A second issue raised by Riddle and Lamb concerns the meaningfulness and utility of the results of our study. Our motivation for conducting this study was our belief that patients with shoulder pain function differently than those without shoulder pain. Differences in the patterns of muscle use, when they exist, must first be identified before their value with regard to assessment and treatment can be realized. Electromyography provides an acceptable and useful method for identifying differences in the sequence, intensity, and timing of muscle activation during functional movement. Our study is unique among the literature related to the shoulder in that indwelling electrodes were used to discern differences between subjects with and without shoulder pathology during the performance of commonly prescribed exercises.

Conservative management for patients with shoulder pain and weakness, with or without signs of instability or impingement, has focused primarily on strengthening the muscles surrounding the shoulder complex. Exercises such as those in our study are chosen simply on the basis of their ability to maximally activate specific shoulder muscles. We agree with the view expressed by Wilk that the effectiveness of therapists' efforts might be enhanced by also addressing the need for improved neuromuscular control during functional movement. As the characteristic patterns of muscle activity in various movements and exercises for injured and uninjured shoulders become better understood, perhaps combining biofeedback and strengthening techniques would provide a more comprehensive retraining program for the patient with shoulder dysfunction.

We acknowledge the caution by Riddle and Lamb to limit direct comparison of the electromyographic results of our study with those of Glousman et al.[5] We maintain, however, that our findings from the infraspinatus muscle during the prone lateral rotation exercise with the arm supported are at least consistent with those of the early cocking phase of throwing, defined by Glousman and colleagues as "a period of abduction and external rotation of the shoulder" ending when contact is made between the forward foot and the ground. No comparison is appropriate between data from the phase of concentric contraction of the prone lateral rotation exercise and data from the late cocking phase, a period consisting mostly of muscle activity to decelerate the upper limb.

Again, we would like to thank Wilk, Riddle, and Lamb for their useful recommendations and constructive criticisms of our work. We also thank Physical Therapy for the opportunity to further elaborate our viewpoints.

Bryon T Ballantyne Sally J O'Hare Jodie L Paschall Mary M Pavia-Smith Angela M Pitz Jerry F Gillon Gary L Soderberg

References

[1] Souza DR, Gross MT. Comparison of vastus medialis obliquus:vastus lateralis muscle integrated electromyographic ratios between healthy subjects and patients with patellofemoral pain. Phys Ther. 1991;71:310-320. [2] Neer CS. Impingement lesions. Clin Orthop 1983; 173:70-77. [3] Hawkins RJ, Kennedy JC. Impingement syndrome in athletes. Am J Sports Med. 1980;8: 151-163. [4] Silliman JF, Hawkins RJ. Current concepts and recent advances in the athlete's shoulder. Clin Sports Med. 1991;10:693-705. [5] Glousman R, Jobe FW, Tibone J, et al. Dynamic electromyographic analysis of the throwing shoulder with glenohumeral instability. J Bone Joint Surg [Am]. 1988;70:220-226.