SHOULDER PAIN IS A VERY common and troublesome

1786 Arthrographic and Clinical Findings in Patients With Hemiplegic Shoulder Pain Sui-Foon Lo, MD, Shu-Ya Chen, MS, PT, Hsiu-Chen Lin, MS, PT, Yick-Fung Jim, MD, Nai-Hsin Meng, MD, Mu-Jung Kao, MD, MHA ABSTRACT. Lo S-F, Chen S-Y, Lin H-C, Jim Y-F, Meng N-H, Kao M-J. Arthrographic and clinical findings in patients with hemiplegic shoulder pain. Arch Phys Med Rehabil 2003; 84:1786-91. Objectives: To identify the etiology of hemiplegic shoulder pain by arthrographic and clinical examinations and to determine the correlation between arthrographic measurements and clinical findings in patients with hemiplegic shoulder pain. Design: Case series. Setting: Medical center of a 1582-bed teaching institution in Taiwan. Participants: Thirty-two consecutive patients with hemiplegic shoulder pain within a 1-year period after first stroke were recruited. Interventions: Not applicable. Main Outcome Measures: Clinical examinations included Brunnstrom stage, muscle spasticity distribution, presence or absence of subluxation and shoulder-hand syndrome, and passive range of motion (PROM) of the shoulder joint. Arthrographic measurements included shoulder joint volume and capsular morphology. Results: Most patients had onset of hemiplegic shoulder pain less than 2 months after stroke. Adhesive capsulitis was the main cause of shoulder pain, with 50% of patients having adhesive capsulitis, 44% having shoulder subluxation, 22% having rotator cuff tears, and 16% having shoulder-hand syndrome. Patients with adhesive capsulitis showed significant restriction of passive shoulder external rotation and abduction and a higher incidence of shoulder-hand syndrome (P.017). Those with irregular capsular margins had significantly longer shoulder pain duration and more restricted passive shoulder flexion (P.017) and abduction (P.020). Patients with shoulder subluxation had significantly larger PROM (flexion, P.007; external rotation, P.001; abduction, P.001; internal rotation, P.027), lower muscle tone (P.001), and lower Brunnstrom stages of the proximal upper extremity (P.025) and of the distal upper extremity (P.001). Muscle spasticity of the upper extremity was slightly negatively correlated with shoulder PROM. Shoulder joint volume was moderately positively correlated with shoulder PROM. Conclusions: After investigating the hemiplegic shoulder joint through clinical and arthrographic examinations, we found that the causes of hemiplegic shoulder pain are complicated. Adhesive capsulitis was the leading cause of shoulder From the Department of Physical Medicine and Rehabilitation (Lo, Meng, Kao); School of Physical Therapy, China Medical University (Chen, Lin); and Department of Radiology (Jim), China Medical University Hospital, Taichung, Taiwan, ROC. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Correspondence to Hsiu-Chen Lin, MS, PT, Sch of Physical Therapy, China Medical University, 91 Shiuesh Rd, Taichung 404, Taiwan, ROC, e-mail: hclin@mail.cmu.edu.tw. 0003-9993/03/8412-7890$30.00/0 doi:10.1016/s0003-9993(03)00408-8 pain, followed by shoulder subluxation. Greater PROM of the shoulder joint, associated with larger joint volume, decreased the occurrence of adhesive capsulitis. Proper physical therapy and cautious handling of stroke patients to preserve shoulder mobility and function during early rehabilitation are important for a good outcome. Key Words: Hemiplegia; Pain; Rehabilitation; Shoulder. 2003 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation SHOULDER PAIN IS A VERY common and troublesome complication after stroke. 1 It has been reported that prevalence of shoulder pain varies from 21% to 72% in stroke patients. 2-4 The variation probably comes from differences in definition of shoulder pain or from differences in recruitment criteria for patients with hemiplegic shoulder pain. Shoulder pain of various causes could delay rehabilitation and could decrease the functional performance of activities of daily living and instrumental activities of daily living. 5,6 For example, hemiplegic patients with shoulder pain need help to eat, dress, drive, and do laundry. The clinical management of hemiplegic shoulder pain usually consists of oral analgesics, intra-articular injection of corticosteroids, physical modalities, and therapeutic exercise. Nonetheless, the most effective treatment protocol has seldom been discussed, most likely because of the uncertainty and variability of the real cause of the shoulder pain. Therefore, defining the etiology of shoulder pain in patients with hemiplegia is meaningful and worthwhile. During the flaccid stage of stroke, proprioceptive impairment, lack of muscle tone, and muscle paralysis hinder the dynamic control and supportive function of the rotator cuff, which causes shoulder subluxation. 5 Consequently, this is a factor contributing to the occurrence of shoulder pain in patients with hemiplegia. 1 Radiologic examination of glenohumeral malalignment in the normal upright position is effective for identifying shoulder subluxation. 7 Clinically, shoulder subluxation can be diagnosed with good reliability by observation and surface palpation of the gap between the acromion and humeral head. 8 In the flaccid shoulder, the stretch force to the surrounding supportive soft tissue results in substantial muscle or tendon tear in the rotator cuff. 6 This is another contributing factor to shoulder pain and range of motion (ROM) limitation. Several physical examinations, such as the drop-arm test or the supraspinatus test, 9 are used to differentiate rotator cuff injury from other shoulder problems. However, these examinations may not be appropriate for patients with hemiplegic shoulders because of the complicated pathology, such as poor muscle strength combined with uncoordinated motor control. 5 In patients with hemiplegia, the definitive diagnosis of rotator cuff tear is usually confirmed by arthrographic examination. 10,11 During stroke recovery, muscle spasticity of the extremities appears in Brunnstrom stage II and lasts until voluntary control

INVESTIGATION OF HEMIPLEGIC SHOULDER PAIN, Lo 1787 Fig 1. Shoulder arthrography showing normal smooth capsular margin and axillary recess (arrows). (B) Shoulder arthrography showing diminished axillary recess (arrows). (C) Shoulder arthrography showing irregular capsular margins (arrows). (D) Shoulder arthrography showing leakage of the contrast medium from glenohumeral joint to subdeltoid bursa, indicating a complete rotator cuff tear (arrows). returns. It is thought that muscle spasticity of the shoulder muscles correlates with shoulder subluxation and ROM limitation, resulting in shoulder pain. 3 Muscle spasticity can be assessed clinically using the Modified Ashworth Scale (MAS), which measures muscle tone distribution during manual movement of the extremity through the ROM 12 for patients with central nervous system lesions. 1,5 Shoulder-hand syndrome occurs in approximately 12% to 25% of stroke patients. 13 It is also called reflex sympathetic dystrophy (RSD), which refers to a complex regional pain syndrome of unknown etiology. In many published studies concerning diagnostic criteria for RSD, pain is the major symptom in the early stages. Distal swelling, vasomotor instability, and dystrophic changes of the skin are common in later stages. 5,14 Early diagnosis and treatment are critical in preventing or minimizing the late changes associated with RSD. RSD can be diagnosed clinically by signs and symptoms. Radionuclide bone scintigraphy supports the diagnosis. 5 Another factor contributing to shoulder pain is adhesive capsulitis (frozen shoulder), 1,5,6 which is characterized clinically by ROM limitation in all directions, with a capsular pattern of restriction. Because the causes of hemiplegic shoulder pain vary, assisted diagnosis with arthrographic techniques is useful and widely used in clinics. Normally, the joint volume exceeds 10mL, the capsular margin is smooth in contour, and an axillary recess is present. The axillary recess is a pouch of the glenohumeral capsule evolving from the inferior rim of the glenoid cavity to the inferior part of the humeral head (fig 1A). 15 Contrast arthrography shows the intra-articular changes of the shoulder joint. 6 Adefinitive diagnosis of adhesive capsulitis can be made if less than 10mL of room air can be injected into the joint during the arthrographic examination. 6,16 Other intra-articular changes related to adhesive capsulitis include diminished or absent axillary recess (fig 1B) and irregular capsular margin (fig 1C). The definitive diagnosis for rotator cuff tear is indicated by contrast medium leakage from the glenohumeral joint to the subdeltoid bursa (fig 1D). The diagnostic accuracy of arthrography for rotator cuff tear has been reported to be as high as 99%. 17 It is regarded as the criterion standard for the diagnosis of rotator cuff tear. 18 Positive contrast arthrography may indicate rotator cuff lesions or confirm the diagnosis of adhesive capsulitis. 6,10,11 Arthrography is a powerful diagnostic tool for evaluating the changes in shoulder joints of patients with hemiplegia. 10,11 Therefore, 1 purpose of our study was to investigate the etiology of shoulder pain and dysfunction in patients with hemiplegia. Another purpose was to determine the correlation between arthrographic measurements and clinical findings in hemiplegic shoulder pain.

1788 INVESTIGATION OF HEMIPLEGIC SHOULDER PAIN, Lo METHODS Thirty-two consecutive patients with hemiplegic shoulder pain after first stroke were recruited for the study. Patients with conditions that predisposed them to shoulder pathology, such as diabetes mellitus, shoulder trauma, cervical disk disease, thyroid disease, and others, were excluded. Our study included clinical physical examinations and shoulder arthrographic studies. The physical examinations were performed by a physiatrist in the physical medicine and rehabilitation department. A radiologist in the radiology department of the same medical center performed the arthrographic tests. Basic anthropometric data (age, sex, body weight, body height), diagnosis, time from stroke to shoulder pain onset, shoulder pain duration, affected side, muscle spasticity, stage of recovery by Brunnstrom stage evaluation, passive range of motion (PROM), active range of motion (AROM), presence or absence of subluxation and shoulder-hand syndrome, and contrast arthrographic findings were recorded. Shoulder subluxation was diagnosed when there was a gap of more than 1 fingerbreadth between the acromion and the head of the humeral bone by palpation. 1,8 Shoulder-hand syndrome was diagnosed clinically by the usual characteristic symptoms (edema, pain at rest and on passive motion of the different joints, bone pain), with involvement of both the hand and the shoulder. 3 The MAS was used to measure the severity of muscle spasticity after stroke and was graded from 0 to 4. 5,12 A grade of 1 was added between grades 1 and 2 to increase the sensitivity of muscle spasticity evaluation. After taking the patient s history, a physiatrist used physical examinations and standard goniometry to evaluate presence or absence of subluxation, muscle spasticity of upper and lower extremities, and ROM of flexion, external rotation, abduction, and internal rotation of shoulder movement. Subsequent shoulder arthrography was conducted by an experienced radiologist. Under fluoroscopic control, 2mL of contrast medium followed by room air were injected into the hemiplegic shoulder joint. Joint volume of less than 10mL indicated adhesive capsulitis, 6,16,19,20 and contrast medium leakage indicated rotator cuff tear. 6,18 The variables analyzed included joint volume, irregular capsular margin, diminished or absent axillary recess, contrast medium leakage, type of shoulder dysfunction, ROM, and muscle spasticity. Temporal variables, such as stroke duration, pain duration, and time from stroke to shoulder pain onset, were recorded and analyzed to identify the possible etiology of hemiplegic shoulder pain. The anthropometric data were evaluated using descriptive analysis. The Student t test was used to examine the variations in ROM measures and temporal variables between patients with hemiplegia with or without arthrographic-related clinical findings. The Mann-Whitney U test was used to differentiate inconsistency in Brunnstrom stages and muscle spasticity among patients with hemiplegia with or without arthrographic and related clinical findings. Pearson correlation coefficients and Spearman correlation coefficients were used for analyzing the correlations between variables. Statistical analyses were performed with the Statistical Package for the Social Sciences, version 8.01. a The significance level was set at.05. RESULTS Among the 32 patients, shoulder pain was aggravated by PROM and AROM, except in patients who could not actively move the affected arm. The patients ages ranged from 44 to 81 years (mean standard deviation [SD], 64.2 10.5y). There were 17 men (53%) and 15 women (47%). All were right handed. Eleven patients (34%) had hemorrhagic stroke, and 21 Fig 2. Distribution of different types of shoulder pain etiology. Abbreviation: FS, frozen shoulder; RCT, rotator cuff tear; SHS, shoulder-hand syndrome. patients (66%) had ischemic stroke. Twelve patients (37%) had right hemiplegia, and 20 patients (63%) had left hemiplegia. Most patients had grade 2 spasticity in the upper extremities (31%) and grade 1 spasticity in the lower extremities (56%). Most motor recovery for upper and lower extremities was Brunnstrom stage III (proximal upper extremity, 38%; distal upper extremity, 50%; lower extremity, 35%). Motor recovery of the upper extremities was stage II or III in about two thirds of the patients and stage IV or V in about one fourth of the patients. In about half the patients, motor recovery of the lower extremities was stage IV or V. In our study, the types of shoulder dysfunction included shoulder-hand syndrome (16%), shoulder subluxation (44%), rotator cuff tear (22%), and frozen shoulder (50%). Half the patients had a joint volume of less than 10mL and diminished axillary recess on arthrographic studies. Four types of shoulder pain cause were recognized, and 11 combinations were found in our study (fig 2). Of the patients, 63% had a single type of shoulder dysfunction, 34% had 2 types of shoulder dysfunction, and 3% had 3 types of shoulder dysfunction. There were 3 patients with idiopathic shoulder pain and dysfunction. The affected side did not correlate with any measured variables of shoulder pain. The shoulder arthrographic test was used to identify the pathology of shoulder dysfunction and to investigate variations in joint volume, irregularity of the capsular margin, diminished axillary recess, and contrast medium leakage. Patients with frozen shoulder tended to have smaller PROM, but the difference was not significant. This implies that other causes, such as soft tissue contracture, might have led to decreased shoulder ROM in the affected upper extremities of our patients. There was no significant difference in stroke duration, time from stroke to pain onset, pain duration, or PROM between patients with and without contrast medium leakage on arthrography (table 1). Most patients could not perform active movement by the time of measurement; therefore, only PROM was analyzed. Patients with shoulder subluxation had significantly greater PROM (table 1), decreased muscle tone (table 2), and lower Brunnstrom stages of proximal upper extremity and distal upper extremity (table 2, fig 3). The hemiplegic patients with frozen shoulder had a significantly higher incidence of shoulder-hand syndrome (table 2). Irregular capsular margin was found only in patients with frozen shoulder, with longer shoul-

INVESTIGATION OF HEMIPLEGIC SHOULDER PAIN, Lo 1789 Table 1: Stroke Duration, Time to Pain Onset, Pain Duration, and ROM in Different Types of Shoulder Dysfunction Irregular Capsular Margin in Frozen Shoulder Rotator Cuff Tear Shoulder Subluxation No (n 8) Yes (n 8) P No (n 25) Yes (n 7) P No (n 18) Yes (n 14) P Stroke duration 3.22 3.44 6.09 3.32.111 4.07 2.83 5.57 4.33.413 5.06 3.68 3.55 2.30.169 Time to pain onset 2.02 2.93 3.09 2.32.429 1.79 1.82 2.96 3.02.205 2.58 2.58 1.35 1.12.081 Pain duration 1.19 0.69 3.00 1.51.008* 2.28 1.48 2.61 2.35.655 2.47 1.57 2.20 1.85.651 PROM flexion 146.25 14.82 123.75 18.27.017* 138.00 20.67 134.29 18.13.670 129.17 18.65 147.50 16.96.007* PROM ER 40.63 10.16 33.75 11.88.234 42.20 16.90 36.43 15.74.425 32.50 12.86 51.79 14.62.001* PROM Abd 111.25 18.08 90.00 14.14.020* 108.20 23.84 104.29 21.49.698 96.11 17.20 121.79 21.98.001* PROM IR 60.63 9.04 50.63 11.78.078 55.00 14.36 60.71 9.32.330 51.67 14.35 62.14 9.94.027* NOTE. Values are mean SD. Abbreviations: Abd, abduction; ER, external rotation; IR, internal rotation. *P.05 denotes statistical significance. der pain duration (table 1), and with more limited ROM in shoulder flexion and abduction (P.02) (fig 4). PROM correlated with joint volume and muscle spasticity. Joint volume was positively correlated with PROM in external rotation and abduction (table 3). Muscle spasticity of the affected upper extremity was moderately negatively correlated with all directions of shoulder joint PROM (table 3). DISCUSSION There were 32 patients with painful hemiplegic shoulder recruited for this study to identify the etiology and pathology of hemiplegic shoulder pain from shoulder arthrographic and clinical findings. The participants in the study were meticulously selected, and patients with other possible causes of shoulder pain were excluded. Careful histories were taken and examinations made to rule out previous problems (eg, diabetes mellitus, shoulder trauma, cervical disk disease, thyroid disease) that might have directly or indirectly caused shoulder pain and dysfunction. The sample selection criteria used in our study aimed to increase the specificity and accuracy of the analyses and conclusions. It was difficult to identify a single pathology of shoulder pain in patients with hemiplegia. In addition to the 4 definite types of etiology, there were several combinations of these types among our patients (fig 2). For example, some patients with adhesive capsulitis also had rotator cuff tears, and some patients with shoulder subluxation also had shoulder-hand syndrome. In clinical practice, the differential diagnosis of hemiplegic shoulder problem is complicated but important. Therefore, the detailed arthrographic and clinical examination should be included to elucidate the true pathology for effective management. Because only 5 stroke patients (16%) were diagnosed with shoulder-hand syndrome, it was impossible to compare variables between subjects with shoulder-hand syndrome and those without shoulder-hand syndrome. Furthermore, among these 5 patients, shoulder pain was the result of a combination of causes, such as RSD combined with subluxation and/or rotator cuff tear. Radionuclide bone scintigraphy was used for diagnosis of RSD. However, the false-negative rate for bone scans can be as high as 40% in patients with shoulder-hand syndrome. 14 Therefore, we did not include shoulder-hand syndrome as an independent variable and could not elucidate the etiology of shoulder-hand syndrome. Muscle spasticity of the upper extremities was categorized as grade 2 on the MAS in most patients. Patients with grade 2 spasticity have limited function for performing daily activities and may require moderate assistance. In most patients ( 66%), motor recovery of the proximal upper extremity was between Brunnstrom stages II and III, and recovery for the distal upper extremity was mostly Brunnstrom stage III. Half the patients showed Brunnstrom stage IV or V for the lower extremities. These findings indicate that the motor performance of the lower extremities was better than that of the upper extremities in our patients, whose average onset duration was 4.4 months. In addition, the mean time to pain onset was 2.0 months. Therefore, in clinical application, intervention should be started earlier than 2 months after stroke, before shoulder problems develop. Table 2: Statistical Results (P values) of Arthrographic Findings and Shoulder Subluxation Frozen Shoulder Irregular Capsular Margin Rotator Cuff Tear Shoulder Subluxation Shoulder-hand syndrome.017*.404.293.433 Brunnstrom stage UE P.926.717.894.025* Brunnstrom stage UE D.341.113 1.000.001* Spasticity UE.616.064.346.001* Abbreviations: D, distal; P, proximal; UE, upper extremity. *P.05 denotes statistical significance. Fig 3. Brunnstrom stage distribution for the patients with and without shoulder subluxation.

1790 INVESTIGATION OF HEMIPLEGIC SHOULDER PAIN, Lo Contrast medium leakage from the glenohumeral joint to the subdeltoid bursa indicates a rotator cuff tear. 6 In our study, patients with contrast medium leakage tended to have longer stroke duration, although this did not reach statistical significance. In the clinical setting, caregivers of patients with hemiplegia often stretch or overstretch the hemiplegic shoulder beyond the normal range to try to regain the normal ROM of the shoulder joint. The normal function of the rotator cuff muscle is to depress the humeral head while flexing the shoulder joint. 6,9 In the flaccid stage, the rotator cuff muscle of a hemiplegic shoulder is weak. Thus, the humeral head cannot glide downward normally, leading to impingement on the shoulder joint and microtrauma. In the spastic stage, abnormal muscle tone results in abnormal scapulohumeral rhythm and contributes to increased glenohumeral friction-compression stress. 5 Incorrect handling of patients results in improper dynamic motor control, and rotator cuff tearing may occur. Thus, it is very important to remind caregivers to depress the humeral head when flexing the shoulder joint and to avoid overstretching the shoulder. In a study by Rizk et al, 21 there were significant decreases in shoulder PROM in the patients with adhesive capsulitis. Our study, however, showed no statistically significant difference in shoulder PROM between patients with and without frozen shoulder. This discrepancy probably results from the different characteristics of the study populations. Decreased shoulder ROM due to other reasons also exists in stroke patients without adhesive capsulitis. Mao et al 19 showed contracture of soft tissues around the shoulder including muscles, tendons, and ligaments due to a prolonged, shortened position. Not all patients with adhesive capsulitis have an irregular capsular margin. Our study shows that the longer the duration of shoulder pain, the greater the possibility of developing an irregular capsular margin. In patients with frozen shoulder, shoulder joint ROM is most restricted in external rotation and abduction. 22 Our results showed that, in addition to external rotation and abduction, the shoulder joint ROM was further limited in flexion in patients with both frozen shoulder and irregular capsular margin. Shoulder pain duration was significantly longer in patients with an irregular capsular margin than in patients without an irregular capsular margin (table 1). Therefore, we suggest preventing the onset of frozen shoulder and treating hemiplegic patients with frozen shoulder as early as possible. Longitudinal follow-up could aid in understanding the natural history of hemiplegic shoulder pain and could produce valuable information for clinical application. Fig 4. Comparison of passive shoulder ROM in patients with adhesive capsulitis with and without irregular capsular margin. Abbreviation: Flex, flexion. Table 3: Correlation Coefficients of PROM, Joint Volume, and Muscle Spasticity Joint Volume Spasticity UE PROM flexion.643* PROM ER.374*.649* PROM Abd.481*.654* PROM IR.418* *P.05 denotes statistical significance. Shoulder joint volume, muscle spasticity, and Brunnstrom stage of the affected distal upper extremity correlated significantly with PROM in different directions. The greater the joint volume, the larger the PROM for external rotation and abduction (table 1). Shoulder joint PROM in all directions was inversely proportional to the muscle spasticity of the upper extremity. Lesions in the upper motoneuron result in muscle tension control dysfunction and in lower motoneuron changes, 23 which can lead to difficulty in maintaining normal ROM. There was no significant correlation between the Brunnstrom stage and the arthrographic findings. This suggests that regardless of the Brunnstrom stage, the possibilities of developing a frozen shoulder or a rotator cuff tear are the same among patients with hemiplegic shoulder pain. The prevalence of rotator cuff tear, diagnosed by the arthrographic finding of contrast medium leakage, was 22% in our study. This percentage is very close to that reported by Hakuno et al in 1984. 11 However, the prevalence of rotator cuff tears reported in another article was much lower. 10 The possible reasons for the differences in reported prevalences are (1) different recruitment criteria (eg, including patients with diabetes mellitus could increase the prevalence of rotator cuff lesions), (2) sampling bias (our method was convenience sampling, because all patients recruited were from a single medical center, and (3) cultural differences (eg, in Taiwan, family members or caregivers prefer to care for patients with hemiplegia with stretching exercises, such as PROM, pulley system, and others). Caregivers place too much emphasis on shoulder ROM, leading to overstretching that can cause rotator cuff injury. Correct handling of patients and supervision by physical therapists during stretching exercises are important to avoid rotator cuff injury in patients with hemiplegia. CONCLUSION The differential diagnosis of hemiplegic shoulder problem relies on clinical and arthrographic evidence. Arthrography has been shown to be a practical method for investigating and diagnosing hemiplegic shoulder problems in a relatively short period. Frozen shoulder was the leading cause of shoulder pain, followed by shoulder subluxation. The longer frozen shoulder lasted, the more shoulder joint PROM was restricted. Shoulder joint ROM in patients with frozen shoulder with irregular capsular margins was further restricted in flexion, in addition to external rotation and abduction. Moreover, there was a substantially high incidence of rotator cuff tear in our study. Therefore, cautious handling of patients and proper physical therapy started early in rehabilitation are important for preventing shoulder problems in patients with hemiplegia. References 1. Gillen G. Upper extremity function and management In: Gillen G, Burkhardt A, editors. Stroke rehabilitation: a function-based approach. St. Louis: Mosby; 1998. p 109-51. 2. Chard MD, Hazleman BL. Shoulder disorders in the elderly (a hospital study). Ann Rheum Dis 1987;46:684-7.

INVESTIGATION OF HEMIPLEGIC SHOULDER PAIN, Lo 1791 3. Van Ouwenaller C, Laplace PM, Chantraine A. Painful shoulder in hemiplegia. Arch Phys Med Rehabil 1986;67:23-5. 4. Bohannon RW, Larkin PA, Smith MB, Horton MG. Shoulder pain in hemiplegia: statistical relationship with five variables. Arch Phys Med Rehabil 1986;67:514-6. 5. O Sullivan SB, Schmitz TJ. Physical rehabilitation: assessment and treatment. 3rd ed. Philadelphia: FA Davis; 1995. p 327-60. 6. Iannotti JP, Williams GR. Disorders of the shoulder: diagnosis and management. Philadelphia: Lippincott Williams & Wilkins; 1999. 7. Shai G, Ring H, Costeff H, Solzi P. Glenohumeral malalignment in the hemiplegia shoulder: an early radiologic sign. Scand J Rehabil Med 1984;16:133-6. 8. Boyd EA, Torrance GM. Clinical measures of shoulder subluxation: their reliability. Can J Public Health 1992;83(Suppl 2): S24-8. 9. Magee DJ. Orthopedic physical therapy. 3rd ed. Philadelphia: WB Saunders; 1997. p 216. 10. Rizk TE, Christopher RP, Pinals RS, Salazar JE, Higgins C. Arthrographic studies in painful hemiplegic shoulders. Arch Phys Med Rehabil 1984;65:254-6. 11. Hakuno A, Sashika H, Ohkawa T, Itoh R. Arthrographic findings in hemiplegic shoulders. Arch Phys Med Rehabil 1984;65:706-11. 12. Bohannon RW, Smith MB. Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther 1987;67:206-7. 13. Tepperman PS, Greyson ND, Hilbert L, Jimenez J, Williams JI. Reflex sympathetic dystrophy in hemiplegia. Arch Phys Med Rehabil 1984;65:442-7. 14. Black-Schaffer RM, Kirsteins AE, Harvey RL. Stroke rehabilitation. 2. Co-morbidities and complications. Arch Phys Med Rehabil 1999;80(5 Suppl 1):S8-16. 15. Vermeulen HM, Obermann WR, Burger BJ, Kok GJ, Rozing PM, van Den Ende CH. End-range mobilization techniques in adhesive capsulitis of the shoulder joint: a multiple-subject case report. Phys Ther 2000;80:1204-13. 16. Jim YF, Tzeng YH, Shen WC. Evaluation of double-contrast shoulder arthrography for post-traumatic shoulder joint pain. Mid Taiwan J Med 1999;4:116-20. 17. Mink JH, Harris E, Rappaport M. Rotator cuff tears: evaluation using double contrast shoulder arthrography. Radiology 1985;157: 621-3. 18. Jim YF, Chang CY, Wu JJ, Chang T. Shoulder impingement syndrome: impingement view and arthrography study based on 100 cases. Skeletal Radiol 1992;21:449-51. 19. Mao CY, Jaw WC, Cheng HC. Frozen shoulder: correlation between the response to physical therapy and follow-up shoulder arthrography. Arch Phys Med Rehabil 1997;78:857-9. 20. Neviaser JS. Adhesive capsulitis and the stiff and painful shoulder. Orthop Clin North Am 1980;11:327-33. 21. Rizk TE, Gavant ML, Pinals RS. Treatment of adhesive capsulitis (frozen shoulder) with arthrographic capsular distension and rupture. Arch Phys Med Rehabil 1994;75:803-7. 22. Kaltenborn FM. Manual mobilization of the joints: the Kaltenborn method of joint examination and treatment. Volume I: the extremities. 5th ed. Oslo: Olaf Norlis Bokhandel; 1999. p 153. 23. Lo SF, Lin CL, Meng NH, et al. Lower motor neuron changes related to upper motor neuron lesion: evidence from electrophysiological study in patients with hemiplegia. Mid Taiwan J Med 1999;4:229-34. Supplier a. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.