Ultrasound and Magnetic Resonance Findings and Correlation in Hemiplegic Patients With Shoulder Pain · Asuman Dogan, ˘ MD,1 I lkay Karabay, MD,1 Cem Hatipoglu, ˘ MD,2 and Nes‚e Özgirgin, MD1 ¹Department of Physical Medicine and Rehabilitation, ²Department of Radiology, Ankara Physical Medicine and Rehabilitation Training and Research Hospital, Ankara, Turkey Background: The aim of this study was to evaluate the ultrasonography (US) and MRI findings in hemiplegic patients with shoulder pain and investigate the correlation between them. It is not possible for these patients to fully perform active range of motion (ROM) and stress tests, so imaging methods take center stage in diagnosis and treatment planning. Materials and Methods: A total of 68 hemiplegic patients with shoulder pain attending the inpatient rehabilitation program were included in the study. MRI and US results from the patient files were recorded. The frequency of each pathology identified by US and MRI was determined. The distribution of MRI and US findings was investigated to see whether there was a statistical difference between the correlation of MRI and US results. Results: The mean (SD) age of the patients was 63.7 (8.3) years and the mean (SD) duration of hemiplegia was 49 (8.9) days. According to the MRI results, glenohumeral and acromioclavicular joint degeneration was found in 77.9% and 79.7% of the patients, respectively; subacromial-subdeltoid bursitis in 80.9%; fluid increase in the joint space in 41.2%; supraspinatus tendinitis in 36.8%; and supraspinatus partial rupture in 33.8%. Shoulder US findings were supraspinatus tendinitis in 54.4%, acromioclavicular joint degeneration in 26.5%, bicipital tendinitis in 20.6%, and subacromial-subdeltoid bursitis in 19.1%. There was a statistically significant difference between MRI and US findings. The results were not compatible with each other (P > .05), and these findings were not consistent with each other since the kappa coefficient was under 0.40 for all these results. Conclusion: Although US is recommended as the first method in determining shoulder pathologies in hemiplegic patients, we suggest that MRI should be used as the first choice in hemiplegic patients with shoulder pain. MRI and US findings were not consistent, and US is dependent on the experience of the operator. MRI should be the first choice in cases where the diagnosis will affect the treatment choice due to the lack of correlation between US and MRI findings. Key words: hemiplegic shoulder pain, ultrasonography, magnetic resonance imaging
S
houlder pain is one of the most common complaints of hemiplegic patients. Shoulder pain affects their daily activities and impairs their quality of life. Determining the causes of shoulder pain and planning the treatment prolong the duration of the hospital stay. Although the reported prevalence rates of shoulder pain in hemiplegia have been stated to be in the range of 38% to 84%, the general rate is thought to be 70%.1 In the hemiplegic painful shoulder, loss of motor control, the severity of paralysis, development of an abnormal movement pattern, glenohumeral subluxation, changes in tone, and secondary changes that occur in the surrounding soft tissue are the mechanisms of action in local problems.2 The complex anatomy and biomechanics of the
Corresponding author: Asuman Do gan, ˘ MD, Ankara, Turkey; phone: 903122559357; e-mail: [email protected]
shoulder and the various mechanisms in hemiplegia require differential diagnosis methods. Direct radiographs, ultrasonography (US), MRI, and arthroscopy are commonly used methods for the diagnosis of local shoulder pathologies in hemiplegic patients. Several methods have been identified in the literature for the differential diagnosis of shoulder pathologies, but their sensitivity, specificity, and predictive values vary. US has been important in the determination of lesions of the rotator cuff and long head of biceps and the evaluation of bursas and joint effusions.3 There is a general consensus on the use of US as the first examination in shoulder pathologies, even though exact sensitivity and specificity could Top Stroke Rehabil 2014;21(Suppl 1):S1–S7 © 2014 Thomas Land Publishers, Inc. www.strokejournal.com doi: 10.1310/tsr21S1–S1
S1
S2
TOPICS IN STROKE REHABILITATION/SUPPL 1, 2014
not be determined from the literature.4-6 The identification of rotator cuff lesions by US and MRI was compared, and they were both found to be sensitive. The specificity was 67% for US and 100% for MRI. The use of US first was recommended for shoulder pathologies.4 Although US is an inexpensive imaging method, it can be affected by the experience of the person who is performing the examination and the features of the device. There are few studies about MRI indications and results in shoulders with hemiplegic pain.7,8 The presence of many etiological factors and the difficulty in performing diagnostic tests during the examination make it challenging to carry out a differential diagnosis for shoulder pain and necessitate the frequent use of imaging techniques in this patient group. The inadequate data provided by radiographic evaluation on soft tissue and the invasiveness of arthroscopy and arthrography lead to a preference for US and MRI in this patient group.9 The frequent occurrence of shoulder pathology such as rotator cuff tears, a similar increase in the incidence of asymptomatic shoulder pathology with age, and the lack of a difference when compared with the hemiplegic asymptomatic shoulder indicate the importance of imaging findings in these patients and the possibility of the discordance arising from the US and MRI results.10 Data in the literature demonstrating that US and MRI findings are not always consistent led us to compare the incidence of shoulder pathologies that were detected with US and MRI and the correlation of the results in our patient group. Materials and Methods Participants
All patients in an inpatient rehabilitation program between May 2007 and January 2009 who had shoulder pain and requested to have US and MRI were included in the study. A total of 126 patients were evaluated. The demographic data, motor level, spasticity, and joint range of motion measurements are recorded into the chart for every hemiplegic patient in our clinic. We also request shoulder x-ray, US, and/or MRI for every patient with shoulder pain. The data are
recorded in the patient chart. Only US and MRI had been requested for some patients, and these patients were excluded (13 patients). Excluded from the study were patients who had hemiplegia for longer than 3 months (12 patients), had a corticosteroid injection in the shoulder within the inpatient period (10 patients), were uncooperative (3 patients), had thalamic pain (3 patients), and had complex regional pain syndrome (17 patients). Procedures
The age, gender, duration of hemiplegia, hemiplegic side, etiology, shoulder pain, motor recovery phase, and muscle tone of the remaining 68 patients were assessed. Pain was assessed with the Visual Analogue Scale (VAS), motor level recovery phases were assessed with the Brunnstrom Scale, and spasticity was assessed with the Ashworth Scale.9,11 Patients were evaluated by x-ray for glenohumeral subluxation. Patients were radiographed for the anteroposterior graphs in the sitting position without supporting the arm.12 The US and MRI results of the patients were obtained from their files, therefore these examinations could have been performed by any radiologist at the radiology unit. The radiologists performing the US were trained and experienced in performing shoulder US. There was no information on whether a special position had been used during the US investigation in the records; it is possible that some US examinations were performed in the supine position, as sitting balance had not fully developed in some patients. For the US, the radiology unit used an ATL 1500 HDI device and a superficial probe. MRI evaluations were performed with the 0.3 TESLA Hitachi Open MR device using the axial T2, oblique coronal STIR, T1 and T2 and sagittal T1 sequences. Statistical analysis
Data analysis was performed with IBM SPSS for Windows, version 11.5 (IBM, Armonk, NY). Continuous data were shown as mean (SD) (minimum – maximum), and number of cases and percentages were used for nominal data. The associations between MRI and US results were analyzed by McNemar test. Diagnostic performance
US and MRI Correlation in Patients With Shoulder Pain
(eg, sensitivity, specificity, positive and negative predictive values, and accuracy) of US, using MRI as the gold standard, was also calculated. Sample size estimation
According to our pilot study, a total sample size of 54 cases achieves 90% power to detect an odds ratio of 5.962 using a 2-sided McNemar test with a significance level of .05. The odds ratio is equivalent to a difference between 2 paired proportions of 0.263, which occurs when the proportion in cell (P12) is 0.316 and the proportion in cell (P21) is 0.053. The proportion of discordant pairs is 0.369. Sample size estimation was performed by using the NCSS and PASS (NCSS LLC, Kaysville, UT) statistical package program. Results The mean age of the patients was 63.7 ± 8.3 years (range, 43-82) with 39 (57.4%) females and 29 (42.6%) males. The hemiplegia was on the right side in 36 (52.9%) and the left side in 32 (47.1%). The hemiplegia type was thromboembolic in 71.9% and hemorrhagic in 28.1%. The mean duration of hemiplegia was 49 ± 8.9 (range, 23-119) days. Median upper extremity tonus values were Ashworth 2 (1-5), Brunstrom 2 (1-6), and degree of subluxation 2 (1-5). Shoulder passive ROM values were flexion 109.2 ± 24.3 (min 60 - max 170), abduction 104.1 ± 25.2, internal rotation 41.5 ± 9.2 (min 20 - max 60), and external rotation 42.7 ± 11.4 (min 20 - max 70). The pain level of the patients was tested using VAS. The pain level was 2.6 ± 0.8 (1-5) at rest and 4.06 ± 2.1 (2-7) after exercise. The MRI and US results of patients are presented in Table 1. MRI most frequently showed glenohumeral and acromioclavicular joint degeneration, subacromial/subdeltoid bursitis, and supraspinatus tendinitis. Acromioclavicular joint degeneration, subacromial/subdeltoid bursitis, supraspinatus tendinitis, and bicipital tendinitis were most common on shoulder US. The results of 15 patients (22.1%) were normal according to the US with no pathology. There was no pathology in 3 patients on MRI (4.4%). The compatibility rates between MRI and US
S3
findings are presented in Table 2. Bicipital tendinitis was the only pathology where a statistically significant difference between the pathology detection rate was not found (P = .648). There was a significant difference between effusion, subacromial bursitis, joint degeneration, and supraspinatus rupture between the modalities (P < .05). These pathologies were more frequent on MRI than US. Only supraspinatus tendinitis was more frequently observed on US than MRI. These findings were not consistent with each other; the kappa coefficient was under 0.40 for all these results. The positive and negative predictive values of the US and MRI results are provided in Table 3. Discussion It is not possible to perform active range of joint motion examinations and some tests showing the continuity of muscles (painful arc, lower arm, Neer and Hawkins, etc) in some patients, because of the motor deficit in shoulders with hemiplegic pain. Therefore, imaging methods may need to be consulted more frequently to demonstrate shoulder pathologies in the hemiplegic patient. Which method should be used? US is stated to be the preferred first method, but its results vary depending on the person performing the US and the device.4-6 Rotator cuff ruptures are seen at a rate of 20% to 40% in the general population. The incidence increases with age, and this was found to be consistent with an increase in the asymptomatic Table 1. Rates of MRI and US findings in the hemiplegic painful shoulder Joint pathologies
MRI, n (%)
US, n (%)
Subacromial bursitis Supraspinatus tendinitis Degeneration, glenohumeral/ acromioclavicular Supraspinatus rupture, partial, complete Bicipital tendinitis Effusion in joint spacing Infraspinatus tendinitis Teres minor tendinitis Subcoracoid bursitis Normal
55 (80.9) 25 (36.8) 53 (77.9)/ 55 (79.7)
13 (19.1) 36 (54.4) 0/18(26.5)
23 (33.8)/2(2.9)
0
11 (16.2) 28 (41.2) 1 (1.5) 0 6 (8.8) 3 (4.4)
14 (20.6) 1 (1.5)
Note: US = ultrasonography.
0 0 15 (22.1)
S4
TOPICS IN STROKE REHABILITATION/SUPPL 1, 2014
Table 2.
Distribution of cases in terms of MRI and US findings MRI Yes
No
Total
Variables
US
n
%
n
%
n
%
P
Kappa
Subacromial subdeltoid bursitis
Yes No Total
12 43 55
17.6 63.2 80.9
1 12 13
1.5 17.6 19.1
13 55 68
19.1 80.9 100.0
S
houlder pain is one of the most common complaints of hemiplegic patients. Shoulder pain affects their daily activities and impairs their quality of life. Determining the causes of shoulder pain and planning the treatment prolong the duration of the hospital stay. Although the reported prevalence rates of shoulder pain in hemiplegia have been stated to be in the range of 38% to 84%, the general rate is thought to be 70%.1 In the hemiplegic painful shoulder, loss of motor control, the severity of paralysis, development of an abnormal movement pattern, glenohumeral subluxation, changes in tone, and secondary changes that occur in the surrounding soft tissue are the mechanisms of action in local problems.2 The complex anatomy and biomechanics of the
Corresponding author: Asuman Do gan, ˘ MD, Ankara, Turkey; phone: 903122559357; e-mail: [email protected]
shoulder and the various mechanisms in hemiplegia require differential diagnosis methods. Direct radiographs, ultrasonography (US), MRI, and arthroscopy are commonly used methods for the diagnosis of local shoulder pathologies in hemiplegic patients. Several methods have been identified in the literature for the differential diagnosis of shoulder pathologies, but their sensitivity, specificity, and predictive values vary. US has been important in the determination of lesions of the rotator cuff and long head of biceps and the evaluation of bursas and joint effusions.3 There is a general consensus on the use of US as the first examination in shoulder pathologies, even though exact sensitivity and specificity could Top Stroke Rehabil 2014;21(Suppl 1):S1–S7 © 2014 Thomas Land Publishers, Inc. www.strokejournal.com doi: 10.1310/tsr21S1–S1
S1
S2
TOPICS IN STROKE REHABILITATION/SUPPL 1, 2014
not be determined from the literature.4-6 The identification of rotator cuff lesions by US and MRI was compared, and they were both found to be sensitive. The specificity was 67% for US and 100% for MRI. The use of US first was recommended for shoulder pathologies.4 Although US is an inexpensive imaging method, it can be affected by the experience of the person who is performing the examination and the features of the device. There are few studies about MRI indications and results in shoulders with hemiplegic pain.7,8 The presence of many etiological factors and the difficulty in performing diagnostic tests during the examination make it challenging to carry out a differential diagnosis for shoulder pain and necessitate the frequent use of imaging techniques in this patient group. The inadequate data provided by radiographic evaluation on soft tissue and the invasiveness of arthroscopy and arthrography lead to a preference for US and MRI in this patient group.9 The frequent occurrence of shoulder pathology such as rotator cuff tears, a similar increase in the incidence of asymptomatic shoulder pathology with age, and the lack of a difference when compared with the hemiplegic asymptomatic shoulder indicate the importance of imaging findings in these patients and the possibility of the discordance arising from the US and MRI results.10 Data in the literature demonstrating that US and MRI findings are not always consistent led us to compare the incidence of shoulder pathologies that were detected with US and MRI and the correlation of the results in our patient group. Materials and Methods Participants
All patients in an inpatient rehabilitation program between May 2007 and January 2009 who had shoulder pain and requested to have US and MRI were included in the study. A total of 126 patients were evaluated. The demographic data, motor level, spasticity, and joint range of motion measurements are recorded into the chart for every hemiplegic patient in our clinic. We also request shoulder x-ray, US, and/or MRI for every patient with shoulder pain. The data are
recorded in the patient chart. Only US and MRI had been requested for some patients, and these patients were excluded (13 patients). Excluded from the study were patients who had hemiplegia for longer than 3 months (12 patients), had a corticosteroid injection in the shoulder within the inpatient period (10 patients), were uncooperative (3 patients), had thalamic pain (3 patients), and had complex regional pain syndrome (17 patients). Procedures
The age, gender, duration of hemiplegia, hemiplegic side, etiology, shoulder pain, motor recovery phase, and muscle tone of the remaining 68 patients were assessed. Pain was assessed with the Visual Analogue Scale (VAS), motor level recovery phases were assessed with the Brunnstrom Scale, and spasticity was assessed with the Ashworth Scale.9,11 Patients were evaluated by x-ray for glenohumeral subluxation. Patients were radiographed for the anteroposterior graphs in the sitting position without supporting the arm.12 The US and MRI results of the patients were obtained from their files, therefore these examinations could have been performed by any radiologist at the radiology unit. The radiologists performing the US were trained and experienced in performing shoulder US. There was no information on whether a special position had been used during the US investigation in the records; it is possible that some US examinations were performed in the supine position, as sitting balance had not fully developed in some patients. For the US, the radiology unit used an ATL 1500 HDI device and a superficial probe. MRI evaluations were performed with the 0.3 TESLA Hitachi Open MR device using the axial T2, oblique coronal STIR, T1 and T2 and sagittal T1 sequences. Statistical analysis
Data analysis was performed with IBM SPSS for Windows, version 11.5 (IBM, Armonk, NY). Continuous data were shown as mean (SD) (minimum – maximum), and number of cases and percentages were used for nominal data. The associations between MRI and US results were analyzed by McNemar test. Diagnostic performance
US and MRI Correlation in Patients With Shoulder Pain
(eg, sensitivity, specificity, positive and negative predictive values, and accuracy) of US, using MRI as the gold standard, was also calculated. Sample size estimation
According to our pilot study, a total sample size of 54 cases achieves 90% power to detect an odds ratio of 5.962 using a 2-sided McNemar test with a significance level of .05. The odds ratio is equivalent to a difference between 2 paired proportions of 0.263, which occurs when the proportion in cell (P12) is 0.316 and the proportion in cell (P21) is 0.053. The proportion of discordant pairs is 0.369. Sample size estimation was performed by using the NCSS and PASS (NCSS LLC, Kaysville, UT) statistical package program. Results The mean age of the patients was 63.7 ± 8.3 years (range, 43-82) with 39 (57.4%) females and 29 (42.6%) males. The hemiplegia was on the right side in 36 (52.9%) and the left side in 32 (47.1%). The hemiplegia type was thromboembolic in 71.9% and hemorrhagic in 28.1%. The mean duration of hemiplegia was 49 ± 8.9 (range, 23-119) days. Median upper extremity tonus values were Ashworth 2 (1-5), Brunstrom 2 (1-6), and degree of subluxation 2 (1-5). Shoulder passive ROM values were flexion 109.2 ± 24.3 (min 60 - max 170), abduction 104.1 ± 25.2, internal rotation 41.5 ± 9.2 (min 20 - max 60), and external rotation 42.7 ± 11.4 (min 20 - max 70). The pain level of the patients was tested using VAS. The pain level was 2.6 ± 0.8 (1-5) at rest and 4.06 ± 2.1 (2-7) after exercise. The MRI and US results of patients are presented in Table 1. MRI most frequently showed glenohumeral and acromioclavicular joint degeneration, subacromial/subdeltoid bursitis, and supraspinatus tendinitis. Acromioclavicular joint degeneration, subacromial/subdeltoid bursitis, supraspinatus tendinitis, and bicipital tendinitis were most common on shoulder US. The results of 15 patients (22.1%) were normal according to the US with no pathology. There was no pathology in 3 patients on MRI (4.4%). The compatibility rates between MRI and US
S3
findings are presented in Table 2. Bicipital tendinitis was the only pathology where a statistically significant difference between the pathology detection rate was not found (P = .648). There was a significant difference between effusion, subacromial bursitis, joint degeneration, and supraspinatus rupture between the modalities (P < .05). These pathologies were more frequent on MRI than US. Only supraspinatus tendinitis was more frequently observed on US than MRI. These findings were not consistent with each other; the kappa coefficient was under 0.40 for all these results. The positive and negative predictive values of the US and MRI results are provided in Table 3. Discussion It is not possible to perform active range of joint motion examinations and some tests showing the continuity of muscles (painful arc, lower arm, Neer and Hawkins, etc) in some patients, because of the motor deficit in shoulders with hemiplegic pain. Therefore, imaging methods may need to be consulted more frequently to demonstrate shoulder pathologies in the hemiplegic patient. Which method should be used? US is stated to be the preferred first method, but its results vary depending on the person performing the US and the device.4-6 Rotator cuff ruptures are seen at a rate of 20% to 40% in the general population. The incidence increases with age, and this was found to be consistent with an increase in the asymptomatic Table 1. Rates of MRI and US findings in the hemiplegic painful shoulder Joint pathologies
MRI, n (%)
US, n (%)
Subacromial bursitis Supraspinatus tendinitis Degeneration, glenohumeral/ acromioclavicular Supraspinatus rupture, partial, complete Bicipital tendinitis Effusion in joint spacing Infraspinatus tendinitis Teres minor tendinitis Subcoracoid bursitis Normal
55 (80.9) 25 (36.8) 53 (77.9)/ 55 (79.7)
13 (19.1) 36 (54.4) 0/18(26.5)
23 (33.8)/2(2.9)
0
11 (16.2) 28 (41.2) 1 (1.5) 0 6 (8.8) 3 (4.4)
14 (20.6) 1 (1.5)
Note: US = ultrasonography.
0 0 15 (22.1)
S4
TOPICS IN STROKE REHABILITATION/SUPPL 1, 2014
Table 2.
Distribution of cases in terms of MRI and US findings MRI Yes
No
Total
Variables
US
n
%
n
%
n
%
P
Kappa
Subacromial subdeltoid bursitis
Yes No Total
12 43 55
17.6 63.2 80.9
1 12 13
1.5 17.6 19.1
13 55 68
19.1 80.9 100.0
