Correlation of 3-dimensional shoulder kinematics to function in subjects with idiopathic loss of shoulder range of motion.

Idiopathic loss of shoulder range of motion (ROM), the loss of shoulder motion without discernable etiology, is a descriptive label for a set of symptoms. (1,2) All patients with idiopathic loss of shoulder RUM complain of decreased motion. (2-10) Additional complaints include disturbed sleep (2-9) and difficulty accomplishing personal hygiene, donning and doffing clothing, (3) and overhead movement, reaching, or rotation activities. (10)

There have been multiple alternative attempts to label this condition. None have been all-inclusive. Codman (2) initially coined the term "frozen shoulder" in 1934. It is the most frequently referred to term labeling the condition found in the literature. (4,5,7-9,11-35) However, usage of the term has not been universal.

Terminology has been based on assumed etiology. Terms based on inflammation include "adhesive capsulitis," (3,10.36-46) "adhesive subacromial bursitis," (7,38,47) "biceps tenosynovitis, (23) "scapulohumeral periarthritis," (1,7,38,47-51) "subdeltoid bursitis," (2,47) "obliterative bursitis," (23) and "tendinitis of the short rotators." (38) Noninflammation-based terms include "stiff and painful shoulder," (2,38,52-54) "calcification of the supraspinatus tendon," (47) "periarticular adhesions," (38) "Duplay disease," (38) "an algodystrophic process," (55) and "checkrein shoulder." (14,56)

Up to 3% of the general population is affected by idiopathic loss of shoulder RUM. (57) Bridgman (48) reported that up to 7% of outpatients seen at a community hospital had symptoms of periarthritis, and Bunker and Anthony (14) reported that more than 5% of all patients in their study who were seen at shoulder clinics were diagnosed with frozen shoulder. Age and sex distributions reported in the literature have been widely variable, with ages ranging from 22 years (43) to 85 years (58) and with the percentage of female subjects ranging from 48% (14) to 84%. (59)

Codman (2) was the first person to publish the opinion that idiopathic loss of shoulder RUM is a self-limiting condition. Grey (17) investigated his belief that frozen shoulder is self-limited. He utilized reassurance, analgesics, and hypnotics as his sole interventions. He concluded that symptoms resolved and full movement returned in 21 of 22 patients within 2 years. Several investigators have presented less favorable prognoses. (4,60)

Binder et al (60) followed 40 patients with frozen shoulder for an average of 44 months. They found that 16 patients still had pain or movement restriction at follow-up and that 5 patients had greater than 25% reduction in total range of movement. Shaffer (4) followed 62 patients for an average of 7 years. Thirty-one patients reported shoulder pain or stiffness, or both, at their final evaluation. Therefore, despite the relatively low percentages of the general population affected by idiopathic loss of shoulder ROM, the long-term limitations experienced by these people suggest that a greater understanding of the condition and more effective intervention approaches are needed.

People with idiopathic loss of shoulder RUM have difficulty completing activities of daily living (ADL). The shoulder RUM necessary to complete certain ADL tasks has been investigated in subjects without shoulder pathology. (61-64) O'Neill et al (62) found that male subjects needed 127.2 degrees of humerus-to-trunk elevation to reach their occiput, 68.7 degrees to reach their opposite shoulder, 86.6 degrees to reach their mouth, and 31.3 degrees to reach their sacrum.

Several investigators have correlated shoulder motion to function. Triffitt (63) utilized Spearman rank correlation coefficients ([r.sub.s]) to determine that humerus-to-trunk elevation ([r.sub.s]=.72) and abduction ([r.sub.s]=.68) had the highest correlation with hair combing in patients seen at shoulder clinics. Internal rotation (IR) ([r.sub.s]=.65) and abduction ([r.sub.s]=.62) had the highest correlation with washing the back. Humerus-to-trunk elevation ([r.sub.s]=.58), IR ([r.sub.s]=.53), and abduction ([r.sub.s]=.55) had the highest correlation with reaching a high shelf. (63) Bostrom et al (64) utilized Spearman rank correlation coefficients to find that humerus-to-trunk flexion and external rotation (ER) correlated moderately well with hand-to-neck ([r.sub.s]=.60 for flexion, [r.sub.s]=.50 for ER), hand-to-opposite shoulder ([r.sub.s]=.59 for flexion, [r.sub.s]=.50 for ER), and hand-behind-back ([r.sub.s]=.52 for flexion, [r.sub.s]=.50 for ER) activity-related tasks in women with rheumatoid arthritis. Twenty-four percent of the variation in personal hygiene ability and 11% of the variation in dressing ability were explained by humerus-to-trunk movement impairment. (64)

Determining the relationship between impairment-based measures such as ROM and patient function is a priority area in physical therapy research. (65) Limitations in ROM are known to be substantial and are the focus of rehabilitation interventions for idiopathic loss of shoulder ROM. However, little is known regarding the relationship between function and ROM limitations or demographic factors for these subjects. Knowledge of this relationship may affect the choice of the most appropriate scientifically based interventions for this condition. Glenohumeral motion was chosen for this investigation as a more accurate representation of the actual joint where the idiopathic motion loss is believed to originate, rather than humerus-to-trunk motion, which is a function of multiple joints. This approach allows interpretation relative to actual glenohumeral joint motion deficits and relationships to function, but does not include compensatory motions that might occur at other shoulder joints.

The purposes of this investigation were: (1) to determine the relationship between glenohumeral ROM and function based on Shoulder Rating Questionnaire (SRQ) scores and (2) to develop a multiple regression equation utilizing glenohumeral ROM and demographic factors to explain functional variation in subjects with idiopathic loss of shoulder ROM.

Method

Subjects

This study was a component of a comprehensive investigation of shoulder ROM and function. (66,67) All subjects consented to participate.

Twenty-one volunteers were investigated. Nineteen subjects were recruited from 4 physical therapy practices and 2 orthopedists. Two subjects were recruited through local newspaper contacts. Fourteen of the 21 subjects had a physician's diagnosis consistent with frozen shoulder or adhesive capsulitis, 4 subjects had rotator cuff-related diagnoses, and 3 subjects had no physician's diagnosis, but all diagnoses and symptoms were considered consistent with frozen shoulder or adhesive capsulitis based on our clinical examination. The subjects were 18 years of age or older. Subjects were screened for inclusion if they had had symptoms for at least 1 month and had not had symptomatic exacerbation over the last month.

Three subjects (14%) were male, and 18 subjects (86%) were female. The subjects' ages ranged from 40 to 67 years ([bar.X]=52.8, SD=6.5). Their height ranged from 1.6 to 1.9 m ([bar.X]=l.7, SD=0.1). Their weight ranged from 61.4 to 86.4 kg ([bar.X]=71.7, SD=9.4). The length of symptoms ranged from 2 to 120 months ([bar.X]=13.6, SD=24.9). Four subjects had experienced idiopathic loss bilaterally. Three subjects reported that their opposite shoulder had returned to normal. The involved shoulder distribution in the subjects with unilateral involvement was 11 right dominant (52%), 4 right nondominant (19%), and 6 left nondominant (29%).

Each subject was examined using a cervical and shoulder clinical screening to eliminate alternative potential causes of loss of shoulder ROM. Specific exclusion criteria are listed in Figure 1. Loss of passive ROM of at least 25% in at least 2 of the following motions was an inclusion criterion: abduction, ER, and IR. Supine passive abduction, ER, and IR side-to-side comparisons were made for the 17 subjects with unilateral involvement. Seven subjects (41%) had at least 25% losses in abduction, ER, and IR; 6 subjects (35%) had losses in abduction and ER; 1 subject (6%) had losses in abduction and IR; and 3 subjects (18%) had losses in ER and IR. The person with current bilateral symptoms had 25% or greater loss in IR as compared with the less involved side.

[FIGURE 1 OMITTED]

Instrumentation

A number of functional scales based on shoulder function are available. (68-72) None were specifically designed to evaluate the shoulder function of subjects with idiopathic loss of shoulder ROM. The SRQ was chosen for this investigation. It contains elements designed to assess global function, pain, ADL, leisure activities, and work-related activities. Subjects with idiopathic loss of shoulder ROM were included in its psychometric properties evaluation. The SRQ has demonstrated acceptable validity, reliability, and responsiveness. (70) Because global glenohumeral function was the overall outcome of interest in this investigation, the total SRQ score was used rather than domain scores. Specific information on the administration and psychometric properties of the SRQ is shown in the Appendix.

Technical advances over the last 20 years allow for accurate and reliable 3-dimensional (3-D) analysis for the evaluation of shoulder ROM. (73-76) Utilizing 3-D kinematic analysis to evaluate scapulohumeral ROM in subjects with idiopathic loss of shoulder ROM has several advantages. It allows for analysis of glenohumeral motion specifically. It is more accurate than goniometry. Finally, errors from evaluating the 3-D movement of the shoulder with a 2-dimensional (2-D) technique are likely.

The Polhemus FASTRAK electromagnetic motion capture system* was used to track the 3-D kinematics of each subject's humerus, scapula, and thorax at a 40-Hz sampling rate per sensor. The FASTRAK system consists of a stationary electromagnetic transmitter and up to four 3-D sensors. An additional sensor attached to a stylus is used to digitally determine the 3-D orientation and position of anatomical landmarks relative to their respective segment sensor. The sensors track the orientation of the segments in reference to the transmitter throughout motion over time.

The manufacturer of the FASTRAK system has reported root-mean-square (RMS) accuracy of 0.15 degree for orientation and 0.3 to 0.8 mm for position within a source-to-sensor separation of 76 cm. (77) Accuracy described as RMS identifies the level of error expected for an average measurement or subject while adjusting for the normal positive and negative distribution of errors by taking the square root of average squared errors. Within-day and within-subject test-retest RMS variability of peak flexion for the primary investigator (PJR) without removing the sensors was less than 1 degree when previously evaluating 4 subjects without shoulder symptoms. (66) Between-day within-subject test-retest RMS variability of peak flexion for the primary investigator was less than 3 degrees when previously evaluating 2 subjects without shoulder symptoms. (66)

Experimental Procedure

Three FASTRAK sensors were used. Each sensor was 2.3 cm in length, 2.8 cm in width, 1.5 cm in height, and weighed 17 g. (77) One sensor was attached to the sternum and one sensor was attached to the skin overlying the flat superior bony surface of the scapular acromion with adhesive tape. The third sensor was attached to a thermoplastic cuff that was secured to the distal humerus with Velcro straps. [dagger] In order to minimize movement error caused by deltoid muscle contraction, the scapular sensor was placed as medially as possible while remaining on the acromion. Figure 2 illustrates the subjects' experimental setup.

[FIGURE 2 OMITTED]

Data collection was performed as outlined in previous investigations. (66,67,78) Subjects were tested while standing. Digitization of bony landmarks on the humerus (lateral epicondyle, medial epicondyle, and a calculated point midway between the lateral and medial epicondyles) scapula (inferior angle, posterior aspect of the acromio-clavicular joint, and root of the spine of the scapula), and thorax (seventh cervical spinous process, eighth thoracic spinous process, suprasternal notch, and xiphoid process) allowed for transformation of sensor data to local anatomically based coordinate systems (75) (Fig. 3).

[FIGURE 3 OMITTED]

Kinematic data were collected for each subject's till active ROM in the following order: flexion, abduction, scapular-plane abduction, ER, and IR. Measurements of ER and IR were collected with the subject's arm at the side (ER1 and IR1) and with the arm as close to 90 degrees of abduction as possible (ER2 and IR2). The scapular plane was defined as 40 degrees anterior to the coronal plane.75 A vertical plywood guide that could be moved to guide flexion, abduction, and scapular-plane abduction was used to keep the subject's arm in the appropriate plane of motion. The scapular plane was marked on the floor for either extremity. The subject remained with the fingertips in light contact on the board while performing the elevation motions.

Data were collected from both shoulders. Data from the involved gleno-humeral joint were used throughout the analysis. The noninvolved shoulder was tested first to allow the subjects to become familiar with the procedure. For the subjects with a history of bilateral involvement, the currently involved shoulder was tested last. Subjects were instructed to move their arm as far as possible for each motion at a self-selected, slow, steady speed (all subjects moved their arm at a rate of less than 1 Hz). Measurements for 5 repetitions of each motion were collected if the subjects were able to perform that number of repetitions. Subjects were allowed to rest, if needed, between sets of motion repetitions. The SRQ was completed after the ROM data collection.

Data Reduction

The digitized anatomical points were the basis for each shoulder segment's clinically relevant local anatomical coordinate system as previously described. (67,75,78) The description of humerus position and orientation related to the scapula and the trunk was calculated through matrix transformations. (75,79) Humerus-to-scapula orientation was described as rotation about y (adduction/ abduction), x' (flexion/extension), and z" (long axis IR/ER) (Cardan angles, Fig. 3). The rotation sequences are consistent with those previously published (75) and allowed for clinically relevant descriptions of humeral motion.

Data Analysis

The intention of this investigation was to use active peak ROM values to investigate the correlation between glenohumeral ROM and function in subjects with idiopathic loss of shoulder ROM. The peak trial may not have been representative of the motion investigated. This concern could be eliminated if there was no difference between trials. Trial effect was analyzed before further analysis was performed.

Trial effect was initially investigated through one-way repeated-measures analyses of variance (ANOVAs) (repeated trials per subject) of the involved and non-involved shoulders in 10 of the subjects. The subgroup consisted of the first and last 5 subjects who had completed the full data collection. If a motion demonstrated a trial effect within the subgroup of subjects, further analysis was done with all subjects who had completed all 5 trials. If a trial effect remained after all available subjects' data were reviewed, a final review of trial effects was performed through standard error of the measurement (SEM) analysis. The SEM was calculated directly as the square root of the within-subjects mean square error from a one-way ANOVA with subjects as the factor, rather than indirectly using intraclass correlation coefficients. (80)

Flexion, abduction, scapular-plane abduction, ER1, IR1, and IR2 did not demonstrate a trial effect. The ER2 demonstrated a trial effect (F ratio=6.42, P=.00) in the subgroup (n=10) analyses. The ER2 trial effect remained in the full analysis (F ratio=6.77, P=.00). Post hoc Bonferroni analyses revealed that repetitions 4 and 5 had greater ER ROM than repetition 1. The SEM for the ER2 data was 2.12 degrees. This magnitude of effect was not judged to be clinically significant. As a result, the peak trial data were used in the remainder of the data analysis.

Two additional potential confounding factors were reviewed as part of the comprehensive investigation of shoulder ROM and function. (66) Neither speed of shoulder movement nor order of data collection was found to be a significant covariate in the peak motion found. It was possible that later tested motions could have been progressively less limited as the shoulder "loosened up." This was not found to be the case.

Descriptive statistics (mean, standard deviation, range) were calculated across subjects for each motion. Potential variables included active peak elevation of the humerus in relation to the scapula and trunk in flexion, scapular-plane abduction, abduction, ER1, IR1, ER2, and IR2.

Pearson product moment correlations (r) and point biserial correlations ([r.sub.pb]) were performed to determine the factors that were strongly associated with the overall SRQ score and the factors that were strongly associated with each other. Factors in the Pearson matrix were the 7 glenohumeral motions and the subject's age, duration of symptoms, height, and weight. Factors in the point biserial matrix were whether the dominant shoulder was involved, whether the subject had a history of bilateral involvement, and whether the subject was taking anti-inflammatory medications. A multiple regression also was calculated to explain the variation in the total SRQ scores.

Several steps were involved in determining which factors were used in the multiple regression. The top 10 factors (of 14 possible) from the Pearson and point biserial correlation matrices were used in the all-possible regression. To avoid multicollinearity, if interfactor correlations were higher than .85, only one of the factors was used in the multiple regression. (81) Variance inflation factors (VIFs) also were assessed in the models to ensure that they remained below a value of 5. (82) Variance inflation factors are direct measures of multicollinearity (the inverse of 1 - the [R.sup.2]). A VIF value greater than 5 indicates an [R.sup.2] greater than .80, and subsequently multicollinearity may be affecting the results. (82)

The sample size limits the number of factors that could be considered. The absolute maximum states there must be at least 2 subjects for each factor in the model. The general rule of thumb is that there should be at least 4 times the number of subjects as predictors. (83) With 21 subjects, a maximum of 10 predictors could be used. Based on the general rule, 4 predictors would be appropriate with 21 subjects. (81) The absolute maximum (2:1) rule was used to determine the 10 factors used in the all-possible regression. The 4:1 rule was utilized to determine the best combination of 4 factors for the multiple regression.

An all-possible regression evaluates all of the possible combinations of factors (allowing for all orders of entry) to determine which factors are significant by themselves or in a group. (83) After accounting for multicollinearity, the 10 most highly correlated factors with the total SRQ score were used in an all-possible regression to determine the most predictive group of 4 factors. Those predictors then were used in the multiple regression, with a forced model ordered as the result of the combination determined by the all-possible regression.

Results

Descriptive Statistics

Means, standard deviations, and ranges of peak humerus motion in relation to the scapula across all involved shoulders for the 7 motions investigated are presented in Table 1. The SRQ scores ranged from 35.5 to 91.2 ([bar.X]=61.9, SD=13.4). Glenohumeral scapular-plane abduction, abduction, flexion, age, ER1, IR1, duration, anti-inflammatory use, bilateral involvement, and dominant involvement were the 10 factors that correlated highest with the total SRQ. Table 2 outlines the Pearson product moment and point biserial correlations. The Pearson product moment correlations between scapularplane abduction and flexion (r=.97, P=.00) and between scapular-plane abduction and abduction (r=.95, P=.00) were greater than .85. Thus, only scapular-plane abduction was used in the subsequent multiple regression analyses, and ER2 and weight also were considered in the all-possible regression. Scapular-plane abduction had the highest individual correlation to the SRQ.

The all-possible regression analysis determined that scapular-plane abduction, ER1, ER2, and weight were the 4 best predictors (Tab. 3). The all-possible regression analysis determined that scapular-plane abduction alone explained the highest individual amount of variation ([R.sup.2]=.38). The addition of ER1 to scapular-plane abduction resulted in the highest 2-factor variation explanation ([R.sup.2]=.41). The addition of ER2 resulted in the highest 3-factor variation explanation ([R.sup.2]=.59). The addition of weight resulted in the overall explanation of 69% of the variance (P=.001, [R.sup.2] =.69) in SRQ scores. All factors were significant predictors individually at alpha <.05.

Discussion

Duration of Symptoms

Codman (2) has stated his belief that idiopathic loss of shoulder ROM is self-limiting. The average duration of subjects' symptoms in this investigation was 13.6 months. All of the subjects continued to demonstrate ROM and functional deficits. Grey (17) found that the frozen shoulder of 21 of 22 subjects spontaneously resolved within 2 years. Although the current investigation was not longitudinal, 2 of the subjects had deficits beyond 2 years after onset of symptoms, supporting the data of other authors (4,60) that deficits can be more long-standing.

Associations of Impairment and Functional Status

No shoulder-specific functional status measures have been specifically validated for subjects with idiopathic loss of shoulder ROM. Although only 5% of the subjects who participated in the development of the SRQ had frozen shoulder, its good overall psychometric properties made it appealing for this investigation. (70)

A correlation of .5 to .75 is generally considered a moderate association, and a correlation over .75 is considered a good-to-strong association. (84) The individual correlations between glenohumeral scapular-plane abduction (r=.62), abduction (r=.56), and flexion (r=.53) and the SRQ met the moderate association threshold. They were moderately associated with the SRQ total score. None of the factors met the strong association threshold (Tab. 2). Additionally, association does not equate to causation.

Multiple regression was utilized to explain variation in SRQ scores. Glenohumeral scapular-plane abduction, ER1, ER2, and weight were found to significantly contribute to the variation in overall SRQ score. Together, these factors explained 69% of the variability. Although these data explain 69% of previously unexplained variation for this population, additional factors that were not included in the model are clearly necessary to fully predict functional status.

Both ER1 and ER2 accounted for part of the variation in the total SRQ score. They are correlated to each other at r=.76, which is less than the recommended multicollinearity cutoff of .85. (81) Different parts of the capsule are tightened when performing ER at different abduction levels. The coracohumeral ligament limits ER at 0 degrees of humerus-to-trunk abduction, (85) whereas the anterior band of the inferior glenohumeral ligament limits ER at 90 degrees of humerus-to-trunk abduction. (86)

Only 1 subject's motion in the current study was less than the values necessary to complete eating activities described by Safaee-Rad et al. (61) This deficit was on the subject's left (nondominant) side. Thirteen of the 21 subjects did not achieve enough scapular-plane abduction to complete reaching the occiput under the parameters outlined by O'Neill et al. (62) This motion is necessary for hair combing.

Shoulder motion is not the only factor involved in completing the activities investigated by Safaee-Rad et al (61) and O'Neill et al. (62) Barker et al (87) investigated the upper-extremity joint coordination strategies utilized to complete ADL tasks. They used a triaxial, flexible electrogoniometer to study the upper limbs of 11 right-hand-dominant subjects through 3 ADL tasks. They determined that it is difficult to determine normal versus abnormal upper-extremity joint coordination strategies. As Barker et al stated, "In fact, the mechanical degrees of freedom available to the upper limb system may imply that such a 'normal' description may always be elusive." (87) (p24)

Although the current study investigated the correlation between active glenohumeral ROM and overall function, Triffitt (63) correlated humerus-to-trunk ROM to specific ADL tasks. The Pearson r correlation of .62 between scapular-plane abduction and the SRQ in our study was higher than Triffitt's Spearman rank correlations between humerus-to-trunk elevation and reaching a high shelf ([r.sub.s]=.58) and lower than his correlation between humerus-to-trunk elevation and hair combing ([r.sub.s]=.72). The correlation between abduction and the SRQ (r=.56) in our study was similar to Triffitt's correlation between abduction and reaching a high shelf ([r.sub.s]=.55) and lower than his correlation between abduction and hair combing ([r.sub.s]=.68) and his correlation between abduction and washing the back ([r.sub.s]=.68). The correlation between IR and the SRQ (r=.22) in our study was much lower than Triffitt's correlation between IR and reaching a high shelf ([r.sub.s]=.53) and his correlation between IR and washing the back ([r.sub.s]=.65). Additionally, IR was not a significant factor in the all-possible or multiple regressions. These findings may suggest different functional deficits or different perceptions of contributions of specific activities to overall function in Triffitt's nonhomogenous group of diagnoses (rotator cuff problems, instability, capsulitis, arthropathy, and others) as compared with the current investigation's people with idiopathic loss of shoulder ROM. A difficulty in comparing the studies directly is that Triffitt utilized 2-D goniometry to investigate humerus-to-trunk motion, whereas we utilized 3-D electromagnetic goniometry to investigate glenohumeral motion. The IR that Triffitt cited may not have been solely glenohumeral.

Bostrom et al (64) correlated humerus-to-trunk ROM and function as measured by the Shoulder Disability Questionnaire and individual functional motions in 63 female subjects with rheumatoid arthritis. The current study's flexion-to-SRQ correlation (r=.53) was similar to their correlations of flexion to hand-to-neck, hand-to-opposite shoulder, and hand-behind-back activity-related tasks. The current study's ER1-to-SRQ correlation (r=.27) was lower than their correlations between ER and hand-to-neck, hand-to-opposite shoulder, and hand-behind-back activity-related tasks. However, the 69% overall SRQ variation explanation in the current study was substantially higher than Bostrom and colleagues' explanation of ROM to personal hygiene and ROM to dressing ability variation (24% and 11%, respectively).

The 3-D investigation of glenohumeral motion should be cautiously compared with the 2-D humerus-to-trunk goniometric techniques that Bostrom et al (64) used. In both our study and the study by Bostrom et al, humeral motion deficits were linked to function; however, Bostrom and colleagues' ER measurement may have included a scapulothoracic component. The larger amount of variation explained in the current investigation may have been due to the glenohumeral analysis that more appropriately investigates the true motion of the shoulder joint. Additionally, idiopathic loss of shoulder ROM may not affect ER as much as rheumatoid arthritis, and rheumatoid arthritis is often bilateral. Most of the subjects in the current study had unilateral involvement. One shoulder may not have caused as much functional deficit.

Limitations

Despite the use of an appropriate experimental setup, skin slip may occur with surface sensor techniques. Skin slip has been shown to be worse as end-range shoulder elevation is approached. (74) Because end-range shoulder elevation in subjects with idiopathic loss of shoulder ROM does not approach normal values, this may not be a large concern for this population. Additionally, end-range ER and IR values may be under-represented with the methods used in this investigation. Previous validation of this technique comparing data collection though surface sensors to bone pin data found a 7.5-degree average error in rotation measurements. (88) Neither ER nor IR had a strong one-factor correlation with the SRQ. Internal rotation was not a significant factor in the multiple regression.

The sample size for this study was small (N=21), because larger sample sizes for this population are difficult to obtain. Such a sample size limited the number of variables that could be considered in the model and did not allow for testing of possible interactions of the independent variables. (83) Furthermore, the coefficients are less stable in regression analyses using small sample sizes. (83,84)

Finally, the investigation may be limited by evaluating only the glenohumeral component of shoulder function. We believe this approach allows for greater interpretation of how the presumed pathology deriving from the glenohumeral joint relates to function. The gleno-humeral joint also is primarily targeted with regard to intervention approaches such as joint mobilization. However, people with idiopathic loss of shoulder ROM also may have deficits in the scapulothoracic component of shoulder motion. Alternatively, compensatory increases in scapulothoracic motion are possible. Inclusion of scapulothoracic data may increase the amount of total SRQ explained.

Clinical Implications

Based on the presumed pathology of the underlying capsular restrictions, a rehabilitation focus on managing ER, IR, and abduction is generally advocated. (13,44,45,89) This is an impairment-based rehabilitation approach. Justification for such an approach relies on an assumption of a relationship between these ROM impairments and a patient's functional status. Results of the current investigation support such an approach, because impairment measures were significantly associated with functional status and 3 ROM impairment factors could explain 59% of the variance in function. Weight was a demographic factor that cannot be changed directly through physical therapy shoulder rehabilitation, yet it explained an additional 10% of the variance in function. Weight was inversely correlated with the SRQ, suggesting that less body weight may lead to better overall self-perception of shoulder function. Although therapists do, at times, make general health recommendations, including weight loss, to their patients, greater clinical attention to this factor for this population may be warranted.

Although the 3-D electromagnetic motion analysis technique used in our study is not a clinical tool, it allows for the most precise and objective measurement of the underlying ROM impairment at the glenohumeral joint. Use of a less precise measure such as goniometry would likely introduce greater random error into the measurements and reduce the strength of association. Use of the more precise technique gives the clinician the best possible knowledge of the relationship of the underlying glenohumeral ROM limitations that he or she is trying to improve and the patient's functional status. The data suggest that an expanded clinical focus on scapular-plane abduction in addition to ER ROM might better improve patient function. This expanded focus is consistent with the premise that scapular-plane abduction is a more functional plane of motion than coronal-plane motions of flexion or abduction. (88) It is important to consider, however, that flexion and abduction were not included in the overall model because of statistical issues of multicollinearity. The univariate data indicate that flexion and abduction are significantly associated with functional status, although not as strongly as scapular-plane abduction. Abduction ROM deficits receive clinical attention as part of the traditionally described "capsular pattern." When considering patient functional goals, the data of our study support assessment and management of flexion deficits as well. The current results may be used to advocate concentration on gleno-humeral elevation, ER1, and ER2 in the treatment of people with idiopathic loss of shoulder ROM. However, because scapulothoracic data are not presented, our data neither support nor refute clinical consideration of the contributions of the scapulothoracic articulation to overall shoulder function.

Research Implications

Although this study provides support for significant associations of ROM impairment and functional status, optimally studies are needed to provide direct evidence of a relationship between improvements in scapular-plane abduction, flexion, and ER and improvements in function. This would require a longitudinal investigation, but would link treatment gains to positive functional improvements.

The SRQ is a self-reported functional status measure. Designing a study that incorporates a functional activity during the collection of ROM data would be insightful. Collection of functional motion data during the investigation would make the motion-to-function analysis more direct. Examples of activities to collect motion data 3-dimensionally include dressing, grooming, and reaching. Collection of data while people perform these activities would allow for more direct comparison with the domains of the SRQ.

Similarly, development of an objective clinical tool that more precisely measures glenohumeral ROM would be beneficial to allow development of more specific interventions for people with idiopathic loss of shoulder ROM. The FASTRAK system is portable and could be used to collect data in the clinic. However, greater ease of use, real-time clinically interpretable data output, and reduced cost are needed prior to such tools being incorporated into clinical practice. Finally, inclusion of scapulothoracic motion in the regression analysis may improve the total SRQ score variation explained. There has been little to no evaluation of the scapulothoracic contribution to idiopathic loss of shoulder ROM thus far in the literature. Scapulothoracic data have been collected as part of the previously completed comprehensive study. (66) Data analysis is ongoing.

Summary and Conclusions

Idiopathic loss of shoulder ROM was the focus of this investigation. Two conclusions can be made from this investigation: (1) active glenohumeral ROM moderately correlates with SRQ-based function in this population, and (2) the combination of glenohumeral scapularplane abduction, ER1, ER2, and weight can be used to predict 69% of overall shoulder function for these patients. Based on the presumed pathology of the underlying capsular restrictions, a rehabilitation focus on managing ER and IR as well as abduction is generally advocated. Results of the current investigation support an additional focus on scapular-plane abduction in order to expand the pathology-based approach toward additional functional goals.

Appendix.

Administration and Psychometric Properties of the Shoulder Rating Questionnaire (SRQ)

The SRQ is self-administered and takes only 5 to 10 minutes to complete. It assesses global function via a 10-point Likert scale of subjects' perception of their shoulder function. It is anchored between 1 ("very poorly") and 10 ("very well"). The remaining domains are scored via a series of multiple-choice questions. The choices ore scaled from 1 ("poorest") to 5 ("best"). There are 4 pain-related questions, 6 activities of daily living-related questions, 3 recreation-related questions, and 4 work-related questions. The average score of the questions for each domain (pain, daily activity, recreation, or work) is multiplied by 2 to determine the domain score. Therefore, each domain's score ranges from 2 to 10 points. L'insalata et al (70) proposed a minimum clinically important difference of 2 points for each domain.

The calculation of the overall score is done through a weighting system. The global assessment raw score is multiplied by 1.5, the pain domain score is multiplied by 4, the daily activities domain score is multiplied by 2, the recreational and athletic activities domain score is multiplied by 1.5, and the work domain score is multiplied by 1. The total score ranges from 17 to 100 points. A higher score is indicative of better function. (70)

The SRQ has shown acceptable validity, reliability, and responsiveness when tested on 73 male and 27 female subjects with diagnoses of impingement syndrome, instability, rotator cuff tear, osteoarthrosis, and adhesive capsulitis. (70) L'insalata et al (70) compared similar domains between the SRQ and the Arthritis Impact Measurement Scales 2 to demonstrate construct validity. Spearman rank-order correlation coefficients relating to pain ranged from .46 for severity of pain at rest to .92 for frequency of pain that interfered with sleep. Activity-related correlations ranged from .71 for combing hair to .8 for reaching a shelf above the head. Work-related correlations ranged from .48 for the inability to work to .89 for needing to modify work conditions.

Internal consistency, evaluated with Cronbach alpha, ranged from .71 for the pain domain to .90 for the daily activities domain. Test-retest domain-based validity, evaluated with Spearman-Brown test-retest analyses, ranged from .94 for the recreational activities domain to .98 for the work-related activities domain. Using L'insulata and colleagues' data, (70) a standard error of the measurement (SEM) of 2.1 points was calculated for the total SRQ score. (a) Finally, L'insalata et al (70) retested 30 subjects who had undergone a successful surgery a year later. All domain and overall scores had improved. The authors documented that the SRQ scores had improved as shoulder function improved. They documented a standardized response mean above 0.8.

(a) Ludewig PM, Borstad JD. Effects of a home exercise programme on shoulder pain and functional status in construction workers. Occup Environ Med. 2003;60: 841-849.

Systemic history affecting either shoulder

Myocardial infarction Osteoporosis Rheumatoid arthritis Stroke

Either shoulder history

Documented rotator cuff tear Manipulation Nonhealed fracture Surgical stabilization At least 2 positive thoracic outlet provocation maneuvers upon examination (a)

Severe skin allergies, sensitivities, or other dermatological problems in the examination area

Any other predisposing event or trauma that resulted in restrictions in range of motion

(a) Walsh MT. Therapist management of thoracic outlet syndrome. J Hand Ther. 1994;7:131-144.

[dagger] Velcro USA Inc, 406 Brown Ave, Manchester, NH 03103.

* Polhemus Inc, 40 Hercules Dr, PO Box 560, Colchester, VT 05446.

References

(1) Duplay S. Dela periarthrite scapulohumerale. Rev Prat D Trav D Med. 1896;53:226. Translation in The Medical Week. 1896;4:253-254.

(2) Codman EA. The Shoulder: Rupture of the Supraspinatus Tendon and Other Lesions In or About the Subacromial Bursa. Boston, Mass: Thomas Todd; 1934:216.

(3) Murnagham JP. Adhesive capsulitis of the shoulder: current concepts and treatment. Orthopedics. 1998;11:153-158.

(4) Shaffer B. Frozen shoulder: a long-term follow-up. J Bone Joint Surg Am. 1992;74:738-746.

(5) Wadsworth CT. Frozen shoulder. Phys Ther. 1986;66:1878-1882.

(6) Bridgman JF. Periarthritis of the shoulder and diabetes mellitus. Ann Rheum Dis. 1972;31:69-71.

(7) Lippmann RK. Frozen shoulder; periarthritis; bicipital tenosynovitis. Arch Sur. 1943;47:283-296.

(8) Miller MD, Wirth MA, Rockwood CA. Thawing the frozen shoulder: the "patient" patient. Orthopedics. 1996;19:849-853.

(9) Reeves B. The natural history of the frozen shoulder syndrome. Scand J Rheumatol, 1975;4:193-196.

(10) Hannafin JA, Chiaia TA. Adhesive capsulitis: a treatment approach. Clin Orthop. 2000;372:95-109.

(11) Andersen NH, Sojberg JO, Johannsen HV, Sneppen O. Frozen shoulder: arthroscopy and manipulation under general anesthesia and early passive motion. J Shoulder Elbow Surg. 1988;7:218-222.

(12) Anton HA. Frozen shoulder. Can Fam Physician. 1993;39:1773-1778.

(13) Bulgen DY, Binder AL, Hazleman BL. Frozen shoulder: a prospective clinical study with an evaluation of three treatment regimes. Ann Rheum Dis. 1984;43:353-360.

(14) Bunker TD, Anthony PP. The pathology of frozen shoulder. J Bone Joint Surg Br. 1995;77:677-683.

(15) Bunker TD. Frozen shoulder: unraveling the enigma. Ann R Coll Surg Engl, 1997;79:210-213.

(16) Fareed DU, Gallivan WR. Office management of frozen shoulder syndrome. Clin Orthop. 1989;242:177-183.

(17) Grey RG. The natural history of "idiopathic" frozen shoulder. J Bone Joint Surg Am. 1978;60:564.

(18) Ha'eri GB, Maitland A. Arthroscopic findings in the frozen shoulder. J Rheumatol. 1981;8:149-152.

(19) Hill JJ, Bogumill H. Manipulation in the treaunent of frozen shoulder. Orthopedics. 1988;11:1255-1260.

(20) Hsu SYC, Chan KM. Arthroscopic distension in the management of frozen shoulder. Int Orthop. 1991;15:79-83.

(21) Lundberg BJ. The frozen shoulder. Acta Orthop Scand Suppl. 1969; 119:1-47.

(22) Mao C, Jaw W, Cheng H. Frozen shoulder: correlation between the response to physical therapy and follow-up shoulder arthrography. Arch Phys Med Rehabil. 1997;78:857-860.

(23) McLaughlin HL. The "frozen shoulder." Clin Orthop. 1961;20: 126-131.

(24) Melzer C, Wallny T, Wirth CJ, Hoffmann S. Frozen shoulder: treatment and results. Arch Orthop Trauma Surg. 1995;114:87-91.

(25) Neviaser RJ, Neviaser TJ. The frozen shoulder: diagnosis and management. Clin Orthop. 1987;223:59-64.

(26) Parker RD, Froimson AI, Winsberg DD, Arsham NZ. Frozen shoulder, part I: chronology, pathogenesis, clinical picture, and treatment. Orthopedics. 1989;12:869-873.

(27) Pearsall AW IV, Osbahr DC, Speer KP. An arthroscopic technique for treating patients with frozen shoulder. Arthroscopy. 1999;15:2-11.

(28) Pollock RG, Duralde XA, Flatow EL, Bigliani LU. The use of arthroscopy in the treatment of resistant frozen shoulder. Clin Orthop. 1994;304:30-36.

(29) Sharma RK, Bejekal RA, Bhan S. Frozen shoulder syndrome. Int Orthop. 1993;17:275-278.

(30) Tamai K, Yamato M. Abnormal synovium in the frozen shoulder: a preliminary report with dynamic magnetic resonance imaging. J Shoulder Elbow Surg. 1997;6:534-543.

(31) Uitvlugt G, Detrisac DA, Johnson LL, et al. Arthroscopic observations before and after manipulation of frozen shoulder. Arthroscopy. 1993;9:181-185.

(32) Warner JJP. Frozen shoulder, diagnosis and management. J Am Acad Orthop Surg. 1997;5:130-140.

(33) Watson L, Dalziel R, Story I. Frozen shoulder: a 12-month clinical outcome trial. J Shoulder Elbow Surg. 2000;9:16-22.

(34) Weber M, Prim J, Bugglin R, et al. Long-term follow up of patients with frozen shoulder after mobilization under anesthesia, with special reference to the rotator cuff. Clin Rheumatol. 1995;14:686-691.

(35) Wiley AM. Arthroscopic appearance of frozen shoulder. Arthroscopy. 1991;7:138-143.

(36) Boyle-Walker KL, Gavard DL, Bietsch E, et al. A profile of patients with adhesive capsulitis. J Hand Ther. 1997;10:222-228.

(37) Emig EW, Schweitzer ME, Karasick D, Lubowitz J. Adhesive capsulifts of the shoulder: MR diagnosis. AJR Am J Roentgenol. 1995;164: 1457-1459.

(38) Neviaser JS. Adhesive capsulitis of the shoulder: a study of the pathological findings in periarthritis of the shoulder. J Bone Joint Surg. 1945;27:211.

(39) Ozaki J, Yoshiyuki N, Sakurai G, et al. Recalcitrant chronic adhesive capsulitis of the shoulder. J Bone Joint Surg Am. 1989;71:1511-1515.

(40) Rizk TE, Christopher RP, Pinais RS. Adhesive capsulitis (frozen shoulder): a new approach to its management. Arch Phys Med Rehabil. 1983;64:29-33.

(41) Rizk TE, Gavant ML, Pinals RS. Treatment of adhesive capsulitis (frozen shoulder) with arthrographic capsular distension and rupture. Arch Phys Med Rehabil. 1994;75:803-807.

(42) Roubal PJ, Dobritt D, Placzek JD. Glenohumeral gliding manipulation following interscalene brachial plexus block in patients with adhesive capsulitis. J Orthop Sports Phys Ther. 1996;24:66-77.

(43) Segmuller HE, Taylor DE, Hogan CS, et al. Arthroscopic treatment of adhesive capsulitis. J Shoulder Elbow Surg. 1995;4:403-408.

(44) Vermeulen HM, Obermann WR, Burger BJ, et al. End-range mobilization techniques in adhesive capsulitis of the shoulder joint: a multiple-subject case report. Phys Ther. 2000;80:1204-1213.

(45) Vermeulen HM, Stokdijk M, Eilers PH, et al. Measurement of three dimensional shoulder movement patterns with an electromagnetic tracking device in patients with a frozen shoulder. Ann Rheum Dis. 2002;61:115-120.

(46) Warner JJP, Answorth A, Marks PH, Wong P. Arthroscopic release for chronic, refractory adhesive capsulitis of the shoulder. J Bone Joint Surg Am. 1996;78:1808-1816.

(47) Dickson JA, Crosby EH. Periarthritis of the shoulder: analysis of 200 cases. JAMA. 1932;99:2252-2257.

(48) Bridgman JF. Periarthritis of the shoulder and diabetes mellitus. Ann Rheum Dis. 1972;31:69-71.

(49) Eto M. Analysis of the scapulo-humeral rhythm for periarthritis scapulohumeralis. J Jpn Orthop Assoc. 1991;65:693-707.

(50) Lee M, Haq AM, Wright V, Longton EB. Periarthritis of the shoulder: a controlled trial of physical therapy. Physiotherapy. 1973;59: 312-315.

(51) Wright V, Haq AM. Periarthritis of the shoulder. Ann Rheum Dis. 1976;35:213-226.

(52) Beaufils P, Prevot N, Boyer T, et al. Arthroscopic release of the glenohumeral joint in shoulder stiffness: a review of 26 cases. Arthroscopy. 1999;15:49-55.

(53) Clarke GR, Willis LA, Fish WW, Nichols PJR. Preliminary studies in measuring range of motion in normal and painful stiff shoulders. Rheumatol Rehabil. 1975;14:39-46.

(54) Loyd JA, Loyd HM. Adhesion capsulitis of the shoulder: arthrographic diagnosis and treatment. South Med J. 1983;76:879-883.

(55) Muller LP, Muller LA, Happ J, Kerschbaumer K. Frozen shoulder: a sympathetic dystrophy? Arch Orthop Trauma Surg. 2000;120:84-87.

(56) Quigley TB. Checkrein shoulder: a type of "frozen" shoulder: diagnosis and treatment by manipulation and ACTH or cortisone. N Engl J Med. 1954;250:188-192.

(57) Ogilvie-Harris DJ, Biggs DJ, Fitsialos DP, MacKay M. The resistant frozen shoulder: manipulation versus arthroscopic release. Clin Orthop. 1995;319:238-248.

(58) Balci N, Balci MK, Tuzuner S. Shoulder adhesive capsulitis and shoulder range of motion in type II diabetes mellitus: association with diabetic complications. J Diabetes Complications. 1999;13:135-140.

(59) Reichmister JP, Friedman SL. Long-term functional results after manipulation of the frozen shoulder. Maryland Medical Journal. 1999; 48:7-11.

(60) Binder AI, Bulgen DY, Hazleman BL, et al. Frozen shoulder: an arthrographic and radionuclear scan assessment. Ann Rheum Dis. 1984;43:365-369.

(61) Safaee-Rad R, Shwedyk E, Quanbury At, Cooper JE. Normal functional range of motion of upper limb joints during performance of three feeding activities. Arch Phys Med Rehabil. 1990;71:505-509.

(62) O'Neill OR, Morrey BJ, Tanaka S, An K. Compensatory motion in the upper extremity after elbow arthrodesis. Clin Orthop. 1992;281: 89-96.

(63) Triffitt PD. The relationship between motion of the shoulder and the stated ability to perform activities of daily living. J Bone Joint Surg Am. 1998;80:41-46.

(64) Bostrom C, Harms-Ringdahl K, Nordemar R. Shoulder, elbow and wrist movement impairment: predictors of disability in female patients with rheumatoid arthritis. Scand J Rehabil Med. 1997;29:223-232.

(65) Clinical research agenda for physical therapy. Phys Ther. 2000;80: 499-513.

(66) Rundquist PJ. Three-Dimeusional Shoulder Kinematics in Subjects With Idiopathic Loss of Shoulder Range of Motion [dissertation], Minneapolis, Minn: University of Minnesota; 2003.

(67) Rundquist PJ, Ludewig PM. Patterns of motion loss in subjects with idiopathic loss of shoulder range of motion. Clin Biomech. 2004;19: 810-818.

(68) Iannotti JP, Williams GR. Disorder of the Shoulder: Diagnosis and Management. Philadelphia, Pa: Lippincott Williams & Wilkins; 1999: 1023-1040.

(69) Leggin BG, Neuman RM, Shaffer MA, et al. Reliability and validity of a shoulder outcome scoring system. J Orthop Sports Phys Ther. 1996;23:65.

(70) L'insalata JC, Warren RF, Cohen SB, et al. A self-administered questionnaire for assessment of symptoms and function of the shoulder. J Bone Joint Surg Am. 1997;79:738-748.

(71) Roach KE, Budiman-Mak E, Songsiridej N, Lertratanakul Y. Development of a shoulder pain and disability index. Arthritis Care Res. 1991;4:143-149.

(72) Stenstrom CH, Arge B, Sundbom A. Home exercise and compliance in inflammatory rheumatic diseases: a prospective clinical trial. J Rheumatol. 1997;24:470-476.

(73) An KN, Browne AO, Korinek S, et al. Three-dimensional kinematics of glenohumeral elevation. J Orthop Res. 1991;9:143-149.

(74) Karduna AR, McClure PW, Michener LA, Sennett B. Dynamic measurements of three dimensional scapular kinematics: a validation study. J Biomech Eng. 2001; 123:184-190.

(75) Ludewig PM, Cook TM. Alterations in shoulder kinematics and associated muscle activity in people with symptoms of shoulder impingement. Phys Ther. 2000;80:276-291.

(76) McQuade KJ, Smidt GL. Dynamic scapulohumeral rhythm: the effects of external resistance during elevation of the arm in the scapular plane. J Orthop Sports Phys Ther. 1998;27:125-131.

(77) 3Space FASTRAK User's Manual, Revision F Colchester, Vt: Polhemus Inc; 1993.

(78) Rundquist PJ, Anderson DD, Guanche CA, Ludewig PM. Shoulder kinematics in subjects with frozen shoulder. Arch Phys Med Rehabil. 2003;84:1473-1479.

(79) Craig JJ. Introduction to Robotics: Mechanics and Control. 2nd ed. New York, NY: Addison-Wesley Co; 1989:5-67.

(80) Fleiss J. The Design and Analysis of Clinical Experiments. New York, NY: John Wiley & Sons Inc; 1986:1-32.

(81) Munro BH. Statistical Methods for Health Care Research. 4th ed. Philadelphia, Pa: JB Lippincott Co; 2001.

(82) Neter J, Wasserman W, Kutner MH. Applied Linear Regression Models. 2nd ed. Homewood, Ill: Richard D Irwin Inc; 1989.

(83) Kleinbaum DG, Kupper LL, Muller KE. Applied Regression Analysis and Other Multivariable Methods. 2nd ed. Boston, Mass: PWS-Kent Publishing; 1988.

(84) Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. Norwalk, Conn: Appleton & Lange; 1993.

(85) Terry GC, Hammon D, France P, Norwood LA. The stabilizing function of passive shoulder restraints. Am J Sports Med. 1991;19:26-34.

(86) O'Brien SJ, Neves MC, Arnoczky SJ, et al. The anatomy and histology of the inferior glenohumeral ligament complex of the shoulder. Am J Sports Med. 1990;18:449-456.

(87) Barker TM, Nicol AC, Kelly IG, Paul JP. Three-dimensional joint co-ordination strategies of the upper limb during functional activities. Proc Inst Mech Eng (H). 1996;210:17-26.

(88) Ludewig PM, Cook TM, Shields RK. Technical note: comparison of surface sensor and bone-fixed measurement of humeral motion. J Appl Biomech. 2002;18:163-170.

(89) Griggs SM, Ahn A, Green A. Idiopathic adhesive capsulitis: a prospective functional outcome study of nonoperative treatment. J Bone Joint Surg Am. 2000;82:1398-1407.

PJ Rundquist, PT, PhD, is Assistant Professor, DPT Program, Krannert School of Physical Therapy, University of Indianapolis, 1400 E Hanna Ave, Indianapolis, IN 46227 (USA) ([e-mail]). Address all correspondence to Dr Rundquist.

PM Ludewig, PT, PhD, is Associate Professor, Program in Physical Therapy, University of Minnesota, Minneapolis, Minn.

Dr Rundquist's contributions included writing, data collection and analysis, project management, and subject recruitment. The manuscript resulted from his dissertation work. Dr Ludewig's contributions included providing facilities and consultation regarding concept, design, data collection and analysis, and writing. The authors acknowledge the members of Dr Rundquist's PhD dissertation committee for their assistance in project development.

The investigation was approved by the Institutional Review Board of the University of Minnesota.

Preliminary data were presented at the Combined Section Meeting of the American Physical Therapy Association; February 12-16, 2003; Tampa, Fla.

This article was received June 18, 2004, and was accepted January 10, 2005.

Table 1.

Descriptive Statistics for Humerus-to-Scapula Peak Motion (in Degrees)
(N=21)

Motion (a)                 [bar.x]   SD    Range

Abduction                    60      18    15-89
Flexion                      72      20    11-102
Scapular-plane abduction     68      18    20-97
ER1                          46      14    17-71
ER2                          54      13    27-76
IR1                           4      11   -28-24
IR2                          -9      12   -42-10

(a) ER1 and IR1=external and internal rotation with subject's arm at
the side, ER2 and IR2=external and internal rotation with subject's arm
as close to 90 degrees of abduction as possible. Data reprinted with
permission from Elsevier from Table 1 (pa ge 814) of. Rundquist PJ,
Ludewig PM. Patterns of motion loss in subjects with idiopathic loss
of shoulder range of motion. Clinical Biomechanics. 2004;19:810-818.

Table 2.
Pearson Product Moment (r) and Point Biserial ([r.sub.Pb]) Correlations
for the Shoulder Rating Questionnaire (SRQ) (N=21) (a)

Factor (b)                 Pearson r    P

Scapular-plane abduction      .62      .00
Abduction                     .56      .01
Flexion                       .53      .01
Age                          -.27      .23
ER1                           .27      .27
IR1                           .22      .33
Duration of symptoms          .21      .35
ER2                           .20      .39
Weight                       -.15      .51
Height                        .10      .67
IR2                          -.07      .76

Factor                   [r.sub.Pb]    P

Anti-inflammatory use       -.40      .10
Bilaterally involved        -.25      .27
Dominant side involved       .30      .32

(a) Motions are for the glenohumeral joint.

(b) ERI and IR1=external and internal rotation with subject's arm at
the side, ER2 and IR2=external and internal rotation with subject's arm
as close to 90 degrees of abduction as possible.

Table 3.
Regression Results (a)

                                              Variance
                                              Inflation   Cumulative
Variable         Coefficient     t       P    Factor      [R.sup.2]

Constant            66.05       3.72   .002
Scapular-plane       0.73       5.49   .000   1.61        .38
 abduction
External             0.90       3.73   .002   3.22        .41
 rotation
 adducted
External            -1.11      -3.77   .002   4.30        .59
 rotation
 abducted
Weight              -0.49      -2.38   .030   1.22        .69

(a) Motions are for the glenohumeral joint.