Hemiplegic shoulder pain is the fourth medical complication after a stroke alongside post stroke depression, falls, and infection especially in urinary tract [23]. The prevalence of hemiplegic shoulder pain in our study was 35.2% while was varying in different studies between 16 and 84% the discrepancy in prevelences between different studies can be attributed to many reasons as pain is a subjective and more difficult to be evaluated and expressed especially in stroke patients with cognitive impairment, aphasia or dysarthria also using different ways for pain definition in different studies (using questionnaires or pain with passive movement) [24,25,26].
The current study revealed that females had higher stroke incidence than males this was agreed with Girijala et al. [27] they found that stroke incidence was common in females than males also shoulder pain had a higher statistically significant in females than males (P = 0.019). Female sex considers one of the most important predictors for development of post stroke shoulder pain in addition to old age and sensory deficit [28]. Westerlind et al. found that gender did not significantly contribute to the final development of post stroke shoulder pain [29], but disagreed with Wang et al. they found stroke was higher incidence in men than women and this was partly explained by several different risk factors between males and females [30].
This study revealed that prevalence of left sided hemiplegia was higher than right sided and those patients with left sided were more prone to develop shoulder pain (P < 0.001). This was agreed with Lindgren et al. they found that left sided hemiparesis considered one of the most important factors to develop shoulder pain in addition to decrease passive shoulder range of abduction and pain frequency [31].
As regard type of stroke this study found that patients with ischemic stroke were statistically significant higher when compared with hemorrhagic stroke (P < 0.001) this was concordant with many studies [28, 32].
The current study revealed that sensory disturbance for light touch was statistically significant higher in patients with shoulder pain versus patients without shoulder pain (P < 0.001) many studies reported that sensory disturbance was one of the most important risk factor for development of hemiplegic shoulder pain [6, 33,34,35].
As regard risk factors for stroke only diabetes mellitus (DM) showed statistically significant effect as risk factor for development of HSP (P = 0.006) this was concordant with Thomas et al. [36] also with Klit et al. [37] they found that DM patients were twice more likely to develop post-stroke shoulder pain versus stroke patients without DM and was attributed to the numerous complications of DM including musculoskeletal particular shoulder pain but other risk factors of stroke like hypertension, ischemic heart disease (IHD), atrial fibrillation (AF), smoking nor dyslipidemia did not exhibit any statistically difference between patients with HSP and without pain.
This research exhibit that shoulder pain was evident in the first 6 months from stroke onset in which the shorter duration of the disease was highly statistically significant when comparing patients with HSP versus patients without HSP (P < 0.001), Poduri [38] found that shoulder pain can occur early post stroke in first 2 weeks but the typical onset was 2–3 months after stroke onset. many factors may contribute to early development of HSP including neurological factors (initial flaccid paralysis, impaired sensation and hemispatial neglect) and factors attributed to shoulder joint (subluxation of the humeral head or rotator cuff injury) [39,40,41,42].
In this study during assessment of daily living activities using Barthel index score we found that patients with HSP had lower score performance when compared with those without hemiplegic shoulder pain this was highly statistically significant lower (P < 0.001). This was consistent with Roy et al. [43] they found that patients with HSP had impaired arm function with poor performance on BI score and showed also a longer hospital stay period in acute stroke.
The current study revealed that stroke patients with shoulder pain had statistically significant lower recovery stages when comparing stroke patients without shoulder pain (P = 0.010) this was in agreement with Yetisgin [44] who found patients with shoulder pain had lower Brunnstrom hand stage when comparing patients without shoulder pain.
This study exhibited that there was no statistically significant difference between hemiplegic patients with and without shoulder pain as regard age, number of risk factors for stroke, muscle power score and Ashworth scale while VAS scale was statistically significant in patients with HSP.
When we compared patients with major dependence and those with moderate dependence on BI score we found that majority of patients with shoulder pain had major dependent with higher statistical significant in females than males (P = 0.031).
In current research we found that there were statistically significantly higher proportion of shorter disease duration, lower muscle power, lower BMR stage and higher proportion of bicipital tendinitis and CRPS in major vs. moderate dependence (P < 0.001, P < 0.001, P < 0.001, P = 0.011 and P = 0.001 respectively).
As regard predictors of the likehood of major versus moderate dependence we found the short disease duration was statistically significant independent predictors on multivariate analysis (P < 0.001) while binary logistic regression analysis on disease duration, muscle power score, sex, nature of stroke (ischemic vs hemorrhagic), sensory disturbance, bicipital tendinitis and complex regional syndrome were statistically significant (P < 0.001, P < 0.001, P = 0.037, P = 0.002, P = 0.002, P = 0.013 and P = 0.006 respectively) on univariate analysis.
In this study shoulder ultrasound and MR imaging were done for 74 patients who suffered from shoulder pain. From 74 patients MR imaging was normal in 19 patients (25.6%) and was abnormal in 55 patients (74.3%) while US was normal in 21 patients (28.3%) and was abnormal in 53 patients (71.6%). The two missed cases by US were diagnosed by MRI as minimal subacromial/subdeltoid bursitis and early adhesive capsulitis.
In the current study the prevalence of gleno-humeral joint effusion, bicipital tendinitis, effusion or tear, subacromial/subdeltoid bursitis, adhesive capsulitis, rotator cuff disorders, shoulder subluxation, and acromio-clavicular joint capsular hypertrophy in painful hemiplegic shoulder were 47.3%, 44.6%, 31%, 29.7, 25.6%, 13.5% and 8% respectively.
Many studies stated that there was high prevalence of bicipital tendinitis, gleno-humeral joint effusion and sub-acromial/sub-deltoid bursitis in patients with hemiplegic shoulder pain [45, 46]. These were matched with current study as we concluded that there was high prevalence of gleno-humeral joint effusion, bicipital tendinitis and sub-acromial/sub-deltoid bursitis in 74 hemiplegic patients with shoulder pain.
In the current study the incidence of adhesive capsulitis was relatively high (29.7%) and this was in agreement with the study of Tavora et al. they concluded that adhesive capsulitis of gleno-humeral joint was significantly higher in hemiplegic shoulder than in other shoulder disorders [47].
In the current study MRI diagnosed 22 patients with adhesive capsulitis while US diagnosed 20 patients with US sensitivity 90.9%, specificity 94.2% and accuracy 93.2%. The two missed cases by US were diagnosed as early adhesive capsulitis by MRI. This was in agreement with the study of Choi and Kim as they reported that MRI could diagnose the increasing of axillary recess capsular thickening and the increasing of T2 signal intensity that occurs in early stage of adhesive capsulitis [21].
In the current study MRI diagnosed 23 case of subacromial/subdeltoid bursitis while US diagnosed 19 cases with high US sensitivity (82.6%) and specificity (98%). The four missed cases by US were diagnosed as minimal subacromial/subdeltoid bursitis by MRI. This was in agreement with the study of El-Shewi et al. they reported that US is less accurate in diagnosis minimal subacromial/subdeltoid bursitis [48]. This study was also in agreement with the study of Zaiton et al. they reported that US had high sensitivity and specificity for diagnosis of subacromial/subdeltoid bursitis (81% and 98%, respectively) [46].
In the current study US diagnosed 31 cases of gleno-humeral joint effusion while MRI diagnosed 35 cases with high US sensitivity (88.5%) and specificity (100%). The four missed cases by US were diagnosed as minimal effusion by MRI. This was in agreement with the studies of Melanie et al. and Mc Nally et al. they reported that US able to discriminate between synovial fluid and the hypoechoic thickened synovium by compression test [49, 50].
In this study there were very good US and MRI agreement for diagnosis of adhesive capsulitis, Subacromial subdeltoid bursitis, bicipital tendinitis, effusion or tear and gleno-humeral joint effusion. This was in agreement with the study of Choi and Kim they reported that there was a good correlation between ability of MRI and US for accurate measurements of axillary recess capsule thickness in patient with adhesive capsulitis [21]. Also this study results were in agreement with study of Zaiton et al. they reported that there was excellent agreement between US and MRI in diagnosis of subacromial/subdeltoid bursitis, diagnosis of bicipital tendinitis, effusion or tear and diagnosis of gleno-humeral joint effusion [46].
In the current study US diagnosed 8 cases of shoulder subluxation while MRI diagnosed 10 cases with 80% US sensitivity, 85.9% US specificity and a moderate US and MRI agreement for diagnosis of shoulder subluxation (kappa = 0.509). This was matched with the study of the Simao et al. they reported that US sensitivity for diagnosis of gleno-humeral joint instability ranged from 20 to 100% and specificity ranged from 25 to 90% [51].
In this study US showed lower sensitivity and specificity than MRI for diagnosis of rotator cuff disorders with a good US and MRI agreement. This was explained in the study of Lenza et al. they concluded that small supraspinatus tendon thickness tears could be missed by US [52].
Our results were matched with the study of El-Shewi et al. they reported the same US sensitivity with our study (83%) for diagnosis of rotator cuff tendinitis and the study of Strobel et al. concluded that US had lower accuracy than MRI in diagnosis of rotator cuff atrophy. [48, 52, 53].
Ii this study US showed the same sensitivity, specificity and accuracy (100%) as MRI in diagnosis of acromio-clavicular joint capsular hypertrophy with perfect agreement (100%) between US and MRI. This was in agreement with the studies of El-Shewi et al. and Melanie et al. as they concluded that dynamic US can directly visualized degenerative changes of acromio-clavicular joint [48, 49].
This study concluded that shoulder US had high accuracy in diagnosis of acromio-clavicular joint capsular hypertrophy, bicipital tendinitis, effusion or tear, Gleno-humeral joint effusion, adhesive capsulitis, and Subacromial/Subdeltoid bursitis. US had moderate accuracy in diagnosis of rotator Cuff disorder and shoulder subluxation.
This study concluded that there were perfect agreement between shoulder US and MRI for diagnosis of acromio-clavicular joint capsular hypertrophy, very good agreement in diagnosis of adhesive capsulitis, subacromial/subdeltoid bursitis, bicipital tendon disorders and gleno-humeral joint effusion, good agreement for diagnosis of rotator cuff disorders and moderate agreement for diagnosis of shoulder subluxation.
US can be used as an alternative or a complementary to MRI for diagnosis of hemiplegic shoulder disorders. The choice between both techniques depends on the clinical condition of the patients, availability of the techniques and the operator US skills. The advantage of US over MRI it is cheaper, not time-consuming, and more applicable for obese patients, critically ill patients and patients who have contraindication for MRI. Also US permit both anatomical and functional assessments of the joint by evaluation of both tissue structure and the dynamic movement of the affected hemiplegic shoulder.
The first limitation of this study was the absence of assessment of depression for the patients enrolled in the study. The second limitation was absence of the correlation between the site and size of the brain insult and their relation with hemiplegic shoulder pain as in this research we studied the local mechanical causes of HSP. Further studies are required to assess the central causes of HSP.