Hemiplegic Shoulder Pain
Good shoulder function is a prerequisite for effective hand function, as well as for performing multiple tasks involving mobility, ambulation, and activities of daily living (ADL). A common sequela of stroke is hemiplegic shoulder pain that can hamper functional recovery and subsequently lead to disability. Poduri reports that hemiplegic shoulder pain can begin as early as 2 weeks poststroke but typically occurs within 2-3 months poststroke. [1, 2] A study by Adey-Wakeling et al found the frequency of poststroke hemiplegic shoulder pain to be 29% over a 12-month follow-up period, with the median pain score being most severe at 4 months. 
Most studies have speculated about the etiology of shoulder pain in hemiplegia but have failed to establish a cause-and-effect relationship. Some of the most frequently suspected factors contributing to shoulder pain include subluxation, contractures, complex regional pain syndrome (CRPS), rotator cuff injury, and spastic muscle imbalance of the glenohumeral joint. [4, 5, 6]
However, identifying the exact mechanism(s) of shoulder pain can be inherently difficult, with many of the current treatment regimens varying according to assumptions made about its cause. Hanger and colleagues suggested it to be highly probable that the cause is multifactorial, with different factors contributing at different stages of recovery (ie, flaccidity contributing to subluxation and subsequent capsular stretch, abnormal tonal and synergy patterns contributing to rotator cuff or scapular instability).  Because of the difficulty in treating shoulder pain once established, early initiation of treatment is valuable.
For individuals who have had strokes with resultant hemiplegia, motor and functional recovery are important steps in the treatment process. Chae and coauthors indicated that the amount of motor recovery is related to the degree of initial severity and the amount of time before voluntary movements are initiated. [8, 9]
Numerous neurofacilitative treatments have been developed in hopes of improving the quality and decreasing the amount of time to recovery. Unfortunately, Chae et al found that the length of stay at most acute inpatient rehabilitation facilities is shortening; they also determined that the primary means of restoring maximal function involves the use of compensatory strategies, rather than the employment of motor control restoration.
In order to understand the pathologic processes and changes that occur in the hemiplegic shoulder, the factors that contribute to normal shoulder position need to be understood. Cailliet has proposed that normal anatomic position involves a well-approximated glenohumeral joint, proper glenoid fossa angle (forward and upward), and proper scapular alignment with the vertebral column. 
The joint is stabilized by musculature (ie, supraspinatus,  deltoid, latissimus) and, to a smaller degree, the shoulder capsule, which supports the humerus. The trapezius, serratus anterior, and rhomboids provide proper scapular alignment. The latissimus also works to depress the scapula.
Erector spinae muscle tone, along with the righting reflex, maintains the vertebral column in an upright alignment. If any of these components are disrupted during the recovery process, then shoulder function may be compromised or a painful shoulder may result.
Pathophysiology of Shoulder Dysfunction
Following a stroke, the brain and body progress through the following series of stages, which Cailliet discusses in detail  :
- Transischemic attack
A gradual progression from one stage to the next usually occurs; however, the stages are not mutually exclusive but instead can occur simultaneously in the affected limb.
Once the inciting injury to the brain occurs, the flaccid stage evolves with a state of areflexia. This stage of areflexia includes loss of muscle tone and volitional motor activity, variable sensory loss, and loss of muscle stretch reflexes.
Muscular support of the humeral head in the glenoid fossa by the supraspinatus and deltoid muscles is lost. This leads to downward and outward subluxation of the humeral head, with the only support coming from the joint capsule.
The shoulder capsule is thin and is composed of 2 tissue layers. The inner synovial layer, the stratum synovium, is highly vascular but poorly innervated, making it insensitive to pain but highly reactive to heat and cold. The outer layer, the stratum fibrosum, is poorly vascularized but richly innervated, predisposing it to pain from stretch. For this reason, Faghri and coauthors suggest that added capsular stretch in a flaccid shoulder may predispose the capsule to irreversible damage and the shoulder to pain. 
Flaccidity of the trapezius, rhomboids, and serratus anterior muscles leads to depression, protraction, and downward rotation of the scapula, which Cailliet believes leads to significant angular changes of the glenoid fossa, subsequently contributing to subluxation.  Also, the spine begins to flex laterally toward the hemiparetic side because of the elimination of the righting reflex, further altering the scapulothoracic relationship.
However, Prevost and colleagues compared the affected and unaffected shoulders by using a 3-dimensional (3-D) radiographic technique that determines the true position of the humeral head in relation to the scapula. This technique revealed less downward rotation of the glenoid fossa than originally expected, and no significant relationship was found between the extent of scapular orientation and the severity of subluxation. [13, 14, 15]
Subsequently, it was concluded that scapular position does not contribute as much to inferior subluxation as was originally thought. Teasell points out that this now appears to be the most widely accepted viewpoint. 
As stroke recovery evolves, flaccidity may progress to spasticity. Cailliet explained that normally, the brainstem contains upper extremity (UE) flexor patterns and lower extremity (LE) extensor patterns that are refined and coordinated by the premotor and neocortexes.  Following a stroke, the connections that control these reflexes can be interrupted, resulting in the release of these basic patterns and the evolution of spasticity and synergy patterns.
If the neurologic deficits become severe enough, primitive tonic neck reflexes may develop. When such neck reflexes are present, the elbow extends when the head turns toward the affected side, and the elbow flexes when the head turns away. The presence of primitive tonic neck reflexes is considered to be prognostically unfavorable for motor recovery.
The first evidence of UE spasticity is internal rotation of the humerus from the subscapularis and pectoralis major; there is debate as to which muscle contributes more strongly to this pattern. The pattern may then progress into the forearm pronators (ie, pronator quadratus, pronator teres, flexor carpi radialis).
Spastic involvement of the rhomboids leads to scapular depression and downward rotation, while the latissimus dorsi contributes to adduction, extension, and internal rotation of the humerus. Biceps brachii spasticity further depresses the head of the humerus and flexes the elbow.
Teasell noted that as spasticity and synergy evolve, there is a failure of the antagonist muscles to relax when the agonist muscles contract, thus creating cocontraction.  For example, during internal rotation, excessive spasticity of the internal rotators of the humerus (ie, subscapularis, pectoralis major, latissimus, teres major) overwhelms the external rotators (ie, supraspinatus, infraspinatus, teres minor).
The muscles causing downward and outward rotation of the scapula, the rhomboids, overwhelm the trapezius and serratus anterior muscles. Spastic, unilateral paraspinal muscles overwhelm those on the contralateral side, causing lateral flexion of the spine toward the affected side.
If neurologic impairment of the completed stroke progresses, synergy patterns, which tend to worsen with initiated efforts, may emerge. Cailliet proposes that the synergy component that usually occurs first is spastic elbow flexion; the shoulder phase is weaker and usually requires a more reflexive status to occur. 
The restrictions created by the synergy patterns create therapeutic challenges to attaining meaningful UE function. UE flexor synergy patterns include the following:
Shoulder/scapular depression (downward rotation and retraction)
Humeral adduction/internal rotation
Forearm pronation (rarely supination)
Wrist/finger flexion (thumb-in-hand position)
When treating patients in flexion synergy, aim therapy at retraining the overwhelmed agonists, stressing the desired components of function, and releasing the uninhibited flexion patterns by initiating opposite movements at the "key points of control."
Epidemiology of Shoulder Pathology
United States statistics
According to Van Ouwenaller and coauthors, shoulder pathology with resulting pain is relatively common in individuals who develop hemiplegia after stroke and/or brain injury.  Van Ouwenaller and colleagues report that shoulder pathology occurs in up to 85% of patients with spastic symptoms and in up to 18% of patients with flaccid symptoms. 
Other clinical trials have reported the general incidence of shoulder pain in patients with hemiplegic stroke to be 16-84%, [17, 18] while that for shoulder subluxation has been found to be as high as 81%. 
Reflex sympathetic dystrophy (RSD) also appears to be a relatively common complication of hemiplegia, with Van Ouwenaller and colleagues reporting an incidence of 27% in patients with spasticity, versus 7% in those with flaccidity.  Other sources have reported an incidence of 12.5-61%.
Consistent with American-based studies, a Turkish study by Aras and coauthors also supported a significant incidence (63.5%) of shoulder pain in stroke patients.  The etiology for this pain also seemed to parallel American studies, with shoulder pain being more frequent in patients with "reflex sympathetic dystrophy, lower motor functional level of shoulder and hand, subluxation, and limitation of external rotation and flexion of shoulder."
Older age and poststroke shoulder pain
A study by Aras and colleagues showed a significantly higher incidence of shoulder pain status poststroke with older age. 
Obtaining an accurate and detailed history is an important part of the examination for hemiplegia-related shoulder problems. For patients who have difficulty with communication, the history may be provided by a family member. Common symptoms of the shoulder and upper extremity (UE) reported by patients with hemiplegia include the following:
Reduced mobility of the shoulder
Pain with movement
The physical examination of a patient with shoulder dysfunction associated with hemiplegia is extensive, because the physician is required to assess the involved musculoskeletal and neurologic conditions. It should include observation, palpation, and neurologic examination.
Observation may reveal the following findings:
Pain with motion
Decreased range of motion (ROM)
On palpation of the shoulder, the following abnormalities may be noted:
Palpable gap between acromion and humeral head (use fingerbreadths or calipers)
To measure subluxation, palpate and/or measure the subacromial space using calipers or a thermoplastic jig. Compare these findings to those of the opposite shoulder. Measure the distance separating the acromial angle and the lateral epicondyle of the humerus using a sliding caliper (anthropometric measure). 
Bohannon and colleagues found that performing shoulder palpation to help diagnose subluxation can be reliably graded, with good interrater reliability and with good correlation with more precise radiographic measurements.  Other authors believe that there are no precise clinical means to measure the degree of subluxation, and if one could be devised, then the benefit of treatment would be validated.
Assess UE pulses. Check the peripheral circulation. Perform the Adson maneuver.
Assess arm function. Perform the Action Research Arm Test.
Evaluation of shoulder pain includes the following:
Ritchie articular index (ordinal measurement)
Shoulder lateral rotation ROM to the point of pain (SROMP) - Precise ratio measurement, requires the use of a goniometer
Evaluate for complex regional pain syndrome (CRPS). For the best prognosis, early recognition and prompt treatment are essential for patients with CRPS. Vasomotor instability (eg, hand edema, MCP tenderness, dystrophic skin changes) should be sought upon examination. If the physical findings are suggestive, evaluation with a triple-phase bone scan that shows periarticular uptake in the wrist and MCP joints of the involved hand can also help with the diagnosis.
The neurologic examination includes the following:
Cognition – Orientation, memory, attention span
Manual muscle testing - Assess strength and tone, evaluate spasticity (modified Ashworth scale)
Sensory evaluation - Light touch, pinprick, vibration, proprioception, 2-point discrimination, stereognosis
Cranial nerves and visual fields
Evaluate for neglect - Letter cancellation test, line bisection test
Evaluate for apraxia
Fugl-Meyer index to test motor performance
Etiology of Poststroke Shoulder Pain
Glenohumeral subluxation basically is defined as a partial or incomplete dislocation that usually stems from changes in the mechanical integrity of the joint. Subluxation is a common problem in patients with hemiplegia, especially during the flaccid stage, and often occurs within 3 weeks poststroke.
Subluxation appears to be caused by the weight of the flaccid arm applying direct mechanical stretch to the joint capsule as well as traction to unsupportive muscles of the shoulder. Teasell suggested that other factors contributing to subluxation include improper positioning, lack of support in the upright position, and pulling on the hemiplegic arm when transferring the patient. 
Controversy exists as to whether there is an association between shoulder subluxation and pain. Subluxation has been a commonly cited cause of shoulder pain and disability, with Yu and coauthors reporting that longitudinal data suggests a correlation between early subluxation and shoulder pain. [22, 23] However, Bohannon and colleagues found no significant correlation between the presence of subluxation and the occurrence of pain, [21, 24] while Wanklyn and coauthors found no association between the severity of subluxation and the degree of pain.  Numerous cases of subluxation without pain have been documented, as have cases of a painful shoulder without subluxation.
A correlation between subluxation and reflex sympathetic dystrophy (RSD) also has been studied. Dursun and coauthors found that subluxation was present in 74.3% of patients with RSD and in 40% of patients without it; of these same patients, 78.6% with subluxation and 38.1% without subluxation reported shoulder pain.  Dursun concluded that shoulder subluxation might be a causative factor of RSD, as well as of shoulder pain.
Spasticity is defined as a velocity-sensitive disorder of motor function causing increased resistance to the passive stretching of muscles and hyperactive muscle stretch reflexes. Following stroke, Teasell reported, supraspinal suppressor areas (pyramidal and extrapyramidal motor systems) that are normally responsible for maintaining the delicate balance between the facilitative and inhibiting influences of alpha and gamma motor neurons are decreased or eliminated, resulting in spasticity, flexor tone, and synergy. 
Van Ouwenaller and colleagues identified spasticity as a prime factor and one of the most common causes of shoulder pain in patients with hemiplegia.  Compared with patients who have flaccidity, patients with spasticity seem to experience a much higher incidence of shoulder pain, which is thought to result from muscle imbalance.
The muscles found to predominate in the shoulder's synergy pattern include the adductors (ie, teres major, latissimus dorsi) and, to a greater extent, the internal rotators (ie, subscapularis, pectoralis major). Bohannon and coauthors reported finding external rotation to correlate most with hemiplegic shoulder pain. [21, 24]
Complex regional pain syndrome
Complex regional pain syndrome (CRPS) is also known as shoulder-hand syndrome, RSD, causalgia, sympathetically maintained pain, Sudeck atrophy, and minor dystrophy. The International Association for the Study of Pain (IASP) has advocated using the terms CRPS type 1 (RSD) and CRPS type 2 (causalgia). The IASP categorization states that RSD develops secondary to noxious stimuli that are not limited to the distribution of a single peripheral nerve, while causalgia starts after a nerve injury.
The incidence of CRPS varies in the literature. Davis and coauthors reported that CRPS occurs in 12.5% of patients who have had a stroke,  while Chalsen and colleagues reported the incidence to be 61%. 
Onset of CRPS is within 3 months poststroke and rarely after 5 months poststroke. In a study, Davis and coworkers demonstrated that of those patients who developed CRPS, 65% had done so by 3 months poststroke, and 98% had done so by 5 months poststroke. 
CRPS is most commonly precipitated by bone or soft-tissue injuries, but in up to 30% of the cases, the patient does not remember experiencing an injury. Snider reported that about 5-8% of patients have an incomplete nerve injury.  Other precipitating factors may include the following:
Upper extremity immobilization
Rotator cuff tear
Glenohumeral joint subluxation
CRPS more commonly affects the upper extremity. Tepperman and colleagues reported metacarpophalangeal (MCP) joint tenderness to be the best diagnostic indicator, having a sensitivity and specificity of 85.7% and 100%, respectively.  Intuitively, however, it is questionable whether any one physical examination maneuver could have such high sensitivity and specificity for a syndrome as complex as RSD.
Using electromyography (EMG), Cheng and Hong found a significant correlation between the presence of spontaneous activity and the development of clinical RSD in 65% of subjects, whereas only 4% of those without spontaneous activity developed RSD.  However, this is not consistent with the definition of RSD set forth by the IASP, which dictates that patients with identifiable nerve lesions may have causalgia but not RSD.
Glenohumeral capsulitis is postulated to play an important role in hemiplegic shoulder pain. Patients usually present with pain and limited passive movement of the shoulder, especially external rotation and abduction.
Joynt reported that adhesive changes may reflect a later stage in the recovery process, when chronic irritation or injury, inflammation, or lack of movement eventually results in adhesions. 
When Rizk and colleagues performed shoulder arthrography in 30 patients with hemiplegic shoulder pain, they found changes in 77% of patients that were consistent only with capsular restriction typical of adhesive capsulitis.  This finding suggests an association between adhesive changes and shoulder pain.
A study by Wanklyn and coauthors found an association between reduced ROM (specifically, external rotation) and hemiplegic shoulder pain, with an incidence as high as 66%.  This association was believed to be due to abnormal muscle tone or structural changes, namely adhesions. Because diminished ROM resulting from either shoulder spasticity or from adhesive capsulitis presents similarly, it is often difficult to distinguish whether pain in the limited hemiplegic shoulder is arising from capsulitis or spasticity, or from a combination of both.
Some patients with hemiplegia complain of lateral shoulder pain that radiates down the arm when moved. This radiating pain seems to correlate with a diagnosis of subacromial bursitis.
Joynt demonstrated that injecting 10 mL of 1% lidocaine into the subjective pain sites related to at least moderate pain relief at the subacromial injection site and improved ROM in 50% of the patients.  This finding suggests that subacromial bursitis can contribute to pain and poor ROM in a significant number of cases.
Brachial plexus traction neuropathies/injury
Patients with hemiplegia who have their flaccid arm in an unsupported, dependent position or patients who have been inappropriately transferred by pulling on the arm, tend to be at increased risk for traction neuropathy.
Wanklyn and coauthors reported a 27% increase in the incidence of shoulder pain in dependent patients after discharge, which may have reflected improper handling at home by caregivers.  For this reason, patient and caregiver education regarding proper transfer techniques and correct handling of the hemiplegic arm should be stressed. Severe sensory loss or neglect also tends to increase the risk for such injuries. Kaplan suggested that plexus injury should be considered in a patient who has atypical return of distal function. 
Heterotopic ossification (HO) presents as calcification of soft tissue around traumatic or neurologically affected joints. Currently, the etiology of HO is unknown.
Patients typically are asymptomatic, and the problem usually is incidentally discovered on radiographs of a joint that is losing ROM.
Clinically, HO can present with local erythema, warmth, induration, and swelling. Cailliet reported that the onset can occur as early as 2 weeks or as late as 3-6 months poststroke. 
Joynt reported that neglect may lead to increased trauma or disturbed perception of the quality of the pain, thereby producing a sensation of pain without the usual pathology.  Snels and coauthors found that on numerous occasions, patients with sensory deficits, visual field deficits, or neglect more commonly experienced recurrent injuries of the shoulder, possibly contributing to capsulitis. 
Thalamic syndrome (also termed central poststroke pain, analgesia dolorosa, Dejerine-Roussy syndrome) usually occurs in less than 5% of stroke survivors, but it is found in 50% of persons who have had a thalamic stroke. The pain can evolve spontaneously or can be evoked by touch, and it is often severe, diffuse, and disabling. Patients describe the pain as burning, tingling ("pins and needles"), sharp, shooting, stabbing, gnawing, dull, or achy. This pain often is refractory to treatment.
The patient also relates experiencing hyperpathia (an exaggerated pain reaction to mild external cutaneous stimulation).
Soft-tissue trauma often is a result of uncontrolled ROM exercises, poor positioning of the hemiplegic patient, or improper transfer technique.
Kumar and colleagues showed that 62% of patients who used an overhead pulley system for therapy and performed ROM exercises experienced shoulder pain irrespective of other pathology, thus demonstrating that overaggressive stretching or ROM should be avoided during the rehabilitation process. 
Patients with poor cognition, neglect, and other sensory deficits tend to be predisposed to traumatic injuries to the affected extremity.
Rotator cuff inflammation/rupture
Because rotator cuff tears are prevalent in the general population, it is often difficult to determine whether a tear was present premorbidly or occurred poststroke.
Through the use of shoulder arthrography, Najenson and coauthors demonstrated the incidence of rotator cuff tear on the affected side to be as high as 40% in patients who were poststroke and were experiencing shoulder pain. The incidence on the unaffected side in these patients was only 16%. 
Other studies, including one by Joynt, have revealed no incidence of rotator cuff tear with hemiplegic shoulder pain.  Teasell reported that hemiplegic shoulder pain is not commonly associated with rotator cuff disorders. 
The differential diagnosis of shoulder pain after stroke includes the following:
Laboratory tests are performed when they are required for a specific indication.
Measure alkaline phosphatase to determine whether heterotopic ossification exists. Obtain rheumatoid factor, antinuclear antibody, and sedimentation rate in patients with suspected rheumatoid arthritis.
The need to objectively measure shoulder subluxation, as well as to determine the effectiveness of slings and other supports used in the treatment and prevention of shoulder subluxation, has led to the development of standardized radiographic techniques. Boyd and colleagues describe that basic radiographic evaluation for shoulder subluxation involves the use of qualitative and quantitative radiographic methods. 
The qualitative method involves visually inspecting the radiographs in order to classify the degree of subluxation. The quantitative method involves comparing the affected shoulder with the unaffected shoulder, or taking a single radiograph of the affected shoulder to measure the amount of subluxation.
Prevost and coauthors proposed a 3-dimensional (3-D) radiographic technique that was shown to be more precise and reliable than other clinical and radiographic techniques in locating the true spatial position of the humeral head relative to the glenoid fossa.  However, since the 3-D technique requires the use of specialized equipment and multiple radiographic exposures, Prevost believes that using one of the more basic 2-dimensional (2-D) techniques is sufficient in assisting with an accurate diagnosis of shoulder subluxation.
Subsequently, Boyd and coauthors proposed their standardized "plane of the scapula method" for classifying subluxation.  This method avoids assumptions about the normality or symmetry of the unaffected shoulder and minimizes the number of radiographs required.
This method has shown moderate measurement validity when comparing the radiograph with the 2 most reliable clinical measures, calipers and fingerbreadths. Even though this technique shows valid correlation with clinical measures and good interrater reliability, it may not be feasible to perform because it requires specialized equipment that is not widely available.
In a study by Kumer et al, diagnostic ultrasonography was used to assess shoulder subluxation in stroke patients and to determine intrarater reliability of these measurements. The acromion-greater tuberosity measurements obtained demonstrated intrarater reliability and discriminate validity. The potential use of ultrasonography to measure subluxation is promising but requires further study. 
A study by Doğun et al indicated that the use of magnetic resonance imaging (MRI) is preferable to that of ultrasonography in evaluating patients with hemiplegia and shoulder pain. The study, of 68 patients, found that ultrasonographic findings with regard to shoulder pathology were not consistent with MRI results, with ultrasonography, for example, demonstrating supraspinatus tendinitis and acromioclavicular joint degeneration in 54.4% and 26.5% of patients, respectively, compared with 36.8% and 79.7% of patients, respectively, on MRI. 
Bone scans assist with the diagnosis of complex regional pain syndrome (CRPS), heterotopic ossification, or other occult etiologies of shoulder pain. Magnetic resonance imaging (MRI) is used when evaluating soft tissue injury.
Electromyography (EMG) may be beneficial when evaluating the following conditions:
If the EMG and nerve studies show an identifiable nerve injury in the distribution of regional symptoms, then according to the International Association for the Study of Pain (IASP) definitions, the patient may have CRPS type-2 (causalgia), but not CRPS type-1 (RSD).
Injections can be used for both diagnosis and therapy. They may help to alleviate shoulder pain and inflammation associated with bursitis and/or tendonitis. Motor point blocks and neurolysis are other procedures that may be used.
Treatment of Poststroke Shoulder Pain
Treatment of subluxation by reduction remains a controversial means of controlling shoulder pain. Slings, arm boards, troughs, and lap trays have not proven to be effective and, in some cases, may result in overcorrection. Sling use also may cause lateral subluxation, impair proprioception, interfere with functional activities, or promote undesirable synergy patterns; furthermore, sling use may not prove beneficial in preventing shoulder subluxation.
Attempts at strapping also have produced variable results. Even though sling use and other supportive devices remain controversial, Yu and coauthors reported that treatment of shoulder subluxation continues to be the standard of care for several reasons, including the following [22, 23] :
Painful shoulder subluxation most commonly is present when the upper extremity is in a dependent position; painful shoulder subluxation improves with joint reduction
Subluxation may have a role in the pathogenesis of other painful conditions by stretching local neurovascular and musculoskeletal tissues
Early prevention is warranted, since chronic shoulder pain often is refractory to treatment
Subluxation may inhibit functional recovery by limiting shoulder range of motion (ROM)
Another form of treatment, neuromuscular electrical stimulation (NMES), has proven moderately successful in the prevention and treatment of subluxation. [40, 41] Yu and colleagues demonstrated substantial reduction in subluxation and, possibly, enhancement of motor recovery and reduction of shoulder pain. [22, 23]
However, it is debated whether NMES should be used prophylactically or whether its use should be held until subluxation develops. (NMES is discussed further under Neuromuscular Electrical Stimulation, below.)
The mainstay of treatment for spasticity begins with physical therapy and the use of ROM and stretching exercises, although overly aggressive stretching should be avoided. Proper positioning also is used as a means of controlling spasticity, by suppressing the evolution of synergy patterns.
Antispasticity medications, as well as casting and orthotics, also should be considered. If conservative treatment fails, then the use of motor point blocks has been advocated as an effective means of improving pain, ROM, and possibly function.
Complex regional pain syndrome
Treatment options are numerous, with physical therapy as the cornerstone. ROM exercises, optimal positioning of the limb, and the avoidance of painful stimuli are all suggested. Other treatments might include nonsteroidal anti-inflammatory drugs (NSAIDs), modalities such as electrical nerve stimulation or ultrasonography), a short course of oral steroids, or a ganglion block.
Treatment of adhesive capsulitis usually involves manual mobilization exercises, analgesics, and possibly steroid injections. If conservative management fails, then the use of distention arthrography or manipulation while the patient is under anesthesia may be indicated.
Early treatment with physical modalities, NSAIDs, steroid injections, and ROM exercises is advocated for the reduction of symptoms and prevention of later complications.
Brachial plexus traction neuropathies
Treatment for traction injuries is limited to the use of supportive care until the return of function.
Treatment begins with ROM exercises, followed by medications (eg, Didronel, Indocin) and irradiation. In severe cases, surgery is necessary to resect the extra-articular bone once it is mature.
Treatment options suggested by Lorish and colleagues include caloric stimulation, prism glasses, visuospatial cueing, computer-assisted training, and compensatory strategies. 
Treatment includes medications, such as analgesics, antidepressants (ie, tricyclic antidepressants), and anticonvulsants. Other treatment alternatives include the administration of a sympathetic blockade and a guanethidine block, as well as the employment of psychological evaluation and treatment. Rarely, surgery is necessary.
The injection of triamcinolone acetonide into the glenohumeral joint has been used to relieve shoulder pain experienced by patients with hemiplegia. Typically, 3 injections of 40 mg of triamcinolone are given via the posterior route. 
Dekker and colleagues demonstrated significant reduction in pain (5 of 7 patients) and improved ROM (4 of 7 patients) that did not reach a level of significance.  None of the secondary outcome parameters (eg, spasticity, motor function, signs and symptoms of shoulder-hand syndrome) showed statistically significant changes either. Dekker concluded that careful positioning, adequate support, and proper handling remain the key actions to prevent hemiplegic shoulder pain.
Another study, a randomized placebo-controlled trial by Snels and coauthors involving intra-articular triamcinolone injections, concluded that treatment effect seemed to decrease shoulder pain and accelerate recovery but was also not found to be statistically significant when compared with placebo. 
Subscapularis motor point nerve block
Many authors, including Wanklyn and colleagues, believe that shoulder pain relates significantly to restriction of external rotation secondary to spasticity.  For this reason, Chironna and Hecht felt that motor block to nerves innervating internal rotators would help relieve the pain caused by internal rotation synergy. 
Using a medial scapular approach, the 2 authors identified the motor points of the nerves to the subscapularis (upper and lower subscapular nerves) via electrical stimulation; then they injected phenol in these points.  An immediate improvement in external rotation, abduction, and flexion was noted, as well as a reduction in pain.
Hecht followed this up with a larger study that showed similar results in pain control and ROM, with the greatest improvement in external rotation. [46, 47] Hecht also reported on a subset of patients with a more spastic pectoralis major than subscapularis. These patients present with greater limitations and pain in abduction and flexion compared with external rotation.
A major complication reported with phenol motor point blocks and neurolysis of mixed nerves is the onset of delayed or chronic neuropathic pain. Fortunately, as reported by Chironna and Hecht, the subscapularis has no sensory nerve component, making the onset of true neuropathic pain unlikely. 
The effect of the block generally lasts from 3-9 months, with the procedure found to be a safe and effective adjunct to conservative treatment. The block is probably most efficacious if performed prior to the development of soft tissue contractures.
Botulinum toxin can be used as a replacement for phenol if the patient does not tolerate the phenol or if the injection is too painful. Botulinum toxin also is preferred when the desired outcome is for slower onset with shorter duration. [48, 49, 50] This procedure is sometimes used when the subscapularis and pectoralis major muscles require nerve block.
Musculocutaneous nerve neurolysis
Elbow flexor spasticity is a common poststroke complication of the flexor synergy pattern. Regular stretching of this muscle group has been suggested as being effective for only a very short period of time before the spastic shortening recurs.
Kong and Chua found that neurolysis (with 50% ethyl alcohol) of the musculocutaneous nerve can significantly improve elbow flexion spasticity and PROM, without affecting strength. These improvements were maintained during the 6-month follow-up period. A small percentage of patients even experienced improved walking balance, decreased finger flexor spasticity, and relief of shoulder pain. The only significant complication reported was a temporary dysesthetic pain in the distribution of the lateral antebrachial cutaneous nerve, a branch of the musculocutaneous nerve. 
Neuromuscular electrical stimulation
The aim of NMES is to reduce subluxation of the hemiplegic shoulder without the use of splints, since no sling design definitively prevents or treats shoulder subluxation.  NMES may even elicit strong sedative effects on pain by acting on sensory nerves. Faghri and coauthors propose that it also could be used prophylactically as a temporary means of splinting the shoulder until recovery of motor function is sufficient to support the glenohumeral joint.  Numerous other studies have suggested that it also improves spasticity and enhances muscle strength of the hemiparetic limb.
Chantraine and colleagues found that patients with hemiplegia and subluxation who received 5 weeks of NMES had significantly more improvement in pain relief, reduced subluxation, quicker motor recovery, and possibly facilitated recovery of shoulder function. These results were maintained for up to 2 years.  However, it was recommended that patients continue exercising to maintain control of their pain.
In patients with chronic hemiplegic stroke and TBI, Yu and coauthors used percutaneous NMES (perc-NMES) in the posterior deltoid and supraspinatus muscles 6 hours a day for 6 weeks.  This resulted in reduced subluxation and improvements in pain and disability. These results were maintained during 3 months of follow-up.
A subsequent study by Yu and coworkers showed that perc-NMES is less painful than transcutaneous NMES, has a much easier application, and has potential for long-term use. This study also found a reduction of shoulder subluxation, with possible enhancement of recovery and improvement in shoulder pain. 
Chae and coauthors reported that intramuscular electric stimulation to the supraspinatus, posterior and middle deltoid, and upper trapezius for 6 hours a day for 6 weeks helped control shoulder pain in chronic stroke survivors.  Compared with the wearing of a cuff-style sling over a similar 6-week time frame, electric stimulation produced better pain control (63% vs 21%), and the effect was maintained through 12 months posttreatment.
At this point, the optimal muscles and number to stimulate has not been established. Yu and coauthors believe that using muscles with strong superior and medially directed forces, as well as those stabilizing the scapula, may significantly enhance the efficacy of this intervention. [22, 23]
Even after 6 months poststroke, forced active repetitive movements of the paretic limb through the use of NMES appears to enhance motor and functional recovery. This has been clinically proven to occur as a result of neuroplasticity, in which active repetitive training of the hemiparetic limb causes functional reorganization in the adjacent intact cortex, subsequently allowing for maximum motor recovery.
Chae and colleagues treated the extensor digitorum communis (EDC) and extensor carpi radialis (ECR) by combining neuromuscular stimulation with active repetitive wrist and finger extension exercises for 1 hour per day for a total of 15 sessions, which produced significantly enhanced motor recovery that was maintained for up to 12 weeks.  However, no significant functional effect was proven.
In patients with hemiplegia, ROM of the shoulder is usually lost early, so Hanger and colleagues recommended that preventive treatments begin as soon as possible, usually within the first 1-2 days poststroke.  Arm support and preservation of joint ROM is performed through early passive motion.
Before active rehabilitation exercises of the extremities are started, Cailliet suggests initiating trunk motions with side-to-side rolling.  As the patient progresses from the supine to the prone position, attempt to maintain the patient in reflex-inhibiting positions.
Gradually implement exercises to raise the arm overhead. Upon regaining the seated position, the patient begins gentle weight-bearing exercises through the impaired arm with the elbow and wrist extended, causing glenohumeral joint reduction and proprioceptive stimulation to the shoulder.
Cailliet has also contended that ROM should be evaluated often because of the almost daily progression or regression of the completed stroke.  Full ROM does not need to be a therapeutic objective but a means for preventing contractures. Also, during passive exercises, the patient should try to assist with motions and hold positions in hopes of encouraging active control of the extremity.
Sensory stimulation, as well as NMES, can be used to initiate sensory-motor reeducation. However, if functional gains plateau because of persistent weakness, then attention may need to focus on functional retraining of the unaffected limb or, through the use of assistive devices, on achieving independence with activities of daily living (ADL). Forced extremity use or constraint therapy also may be considered.
A major goal of early stroke management is the prevention of muscle spasticity that could interfere with the patient's potential for regaining function. As muscle tone returns to the hemiplegic limb, spasticity may progressively increase. Carr and Kenney proposed the use of reflex-inhibiting postures that tend to discourage the development of spasticity, contractures, and other undesirable sequela. 
Even with proper positioning, spasticity may evolve, thus requiring frequent slow stretching, along with the use of splints, to help reduce tone. Overly aggressive stretching should be avoided since it can have a deleterious effect on the treated shoulder by inducing a worsened synergy.
Development of motor control
As hemiparetic limb movements evolve, they show a combination of hypertonicity and weakness, features typical of an upper motor neuron lesion. The recruitment patterns of individual motor units in these affected muscles are slow and inconsistent. Brandstater has related that the variable degrees of cocontraction of the agonist and antagonist muscle groups cause movements to be slow and clumsy.  Because of the importance in coordinating these movements during recovery, multiple approaches have been developed in an attempt to improve functional outcome.
More conventional rehabilitation methods involve reeducating weak muscles by strengthening and stretching. But because these methods have produced marginal results, other techniques that attempt to counter the evolution of normal pathological processes and encourage the use of sensory inputs to facilitate muscle activity have been developed.
Developed by the Bobaths for the treatment of cerebral palsy, the neurodevelopmental technique (NDT) is probably the most widely accepted method used in the development of motor control in patients with hemiplegia. According to Brennan, exercises that promote normal muscle tone and diminish excessive spasticity through the use of reflex-inhibiting postures allow the patient to feel normal movements while preventing the use of compensatory motions.  As Lorish and colleagues indicate, this facilitates higher-level reactions and patterns in order to attain normal automatic motor responses that eventually allow the performance of skilled voluntary movement. 
Brandstater suggests that reciprocal inhibition also be used to temporarily reduce tone in spastic antagonist muscles through the use of a vibratory stimulus. 
The sensory integration system, as advocated by Rood and described by Brandstater, involves superficial sensory stimulation and feedback to the affected extremity by means of brushing, stroking, tapping, icing, vibration, sudden or gentle stretching of the muscle, and even electrical stimulation to facilitate muscle activation. 
The use of robot-aided sensorimotor stimulation also has been implemented. Volpe and coauthors researched the effects of using a robotic device that interacts with the patient in real-time to enhance motor outcome.  The robot was able to guide the powerless limb and provided a sensorimotor experience that responds quickly, just like hand-over-hand therapy. In their randomized blinded study, robot-trained subjects demonstrated improved motor outcome of the shoulder and elbow, as well as improved function.
Theoretically, if motor recovery does in fact depend on motor relearning, then optimal therapies can be tailored for individual patient needs through treatments performed by robotic devices. Overall, Volpe believes that "focused sensorimotor exercise appears to produce better motor outcome." 
Functional utilization of evolving synergies
Assuming normal stages of recovery following stroke, Brunnstrom encouraged reflex tensing in order to develop flexor and extensor synergies during early recovery. According to Reding, induced synergistic reflexes transition into voluntary activation through central facilitation when applied to physiotherapy.  Functional utilization uses techniques such as tonic stretches and voice commands to elicit muscle contractions.
Motor relearning program
Developed by Carr and Shephard, this practical method emphasizes motor relearning by practicing task-specific motor activities while sitting, standing, or walking.  Therapists analyze each task, determine which components the patient cannot perform or has difficulty performing, train the patient in those components of the task, and ensure carryover of this training during daily activities.
Brennan has maintained that ultimately, treatment focuses on eliminating unnecessary muscle activity, subsequently expediting skilled motor activities.  Lorish and colleagues have contended that the use of task-specific training programs tends to be more consistent with modern theories of motor relearning. 
Biofeedback is based on muscular relaxation and/or reeducation by verbal, visual, sensory, or auditory responses. Biofeedback is used in an attempt to relax the antagonist muscles, subsequently allowing the opposed agonists to function more effectively. In order to reeducate the upper extremity, the spastic scapular and glenohumeral antagonist muscles need to be released in order for the agonists to work more proficiently.
A common type of biofeedback, which was first introduced in 1960, involves the use of EMG for neuromuscular reeducation. Overall, trials involving EMG biofeedback have shown mixed results, and its cost-effectiveness is uncertain. However, a meta-analysis by Schleenbaker and Mainous showed it to be an effective tool for neuromuscular reeducation and improving functional outcomes in stroke patients with hemiplegia. 
Proprioceptive neuromuscular facilitation
Developed by Kabat, Knott, and Voss, proprioceptive neuromuscular facilitation (PNF) involves repeated muscle activation of the limbs by quick stretching, traction, approximation, and maximal manual resistance in functional directions (ie, spiral and diagonal patterns) to assist with motor relearning and increasing sensory input.
Brennan asserts that PNF is based on the principles of normal human development (ie, mass movements precede individual movements, reflexive movements precede volitional movements, developments occur cephalically to caudally, control is gained proximally prior to distally, the timing of normal movements is distal to proximal).  Lorish and coauthors have considered it to be an optimal method of stretching in patients with hemiplegia. 
In an attempt to relax spastic antagonist muscle groups, rhythmic stabilization can be used, which involves alternating voluntary contractions of agonist and antagonist muscles. However, Brandstater found PNF to be more effective when muscle weakness is not due to upper motor neuron lesions. 
Chae and colleagues reported that active repetition can maximize motor relearning when used in the appropriate candidate. [8, 9] Parry and coworkers found that stroke patients who were less severely impaired (ie, possessed some early volitional arm movement) prior to treatment benefited from the use of early additional therapies that involved repetitive movements and functional tasks.  However, patients with severe arm impairment showed very little improvement in function irrespective of receiving additional therapies.
These data support previous clinical trials that suggest there is no current physical therapy approach that results in sustained improvements of upper limb function in patients who are severely impaired. In patients who are severely impaired, the use of adaptive techniques and equipment may be an appropriate rehabilitation strategy.
Blended approach to physical therapy
As of yet, numerous clinical trials have not proven that application of any one facilitative approaches improves patient outcome over conventional therapy. [61, 62, 63, 64, 65, 66] They have also not yet proven that any of these approaches is clearly superior to the others. [8, 42, 54]
Currently, common clinical practice involves implementing elements of various techniques, with Cailliet suggesting that the following basic concepts be used during muscle reeducation  :
Patient should visualize (ie, mirror) specific movements
Verbally reinforce intended movements and encourage the feel of specific motions
Copy similar motions performed simultaneously by the contralateral arm
Position the upper extremity to decrease scapular depression and retraction
Apply sensory stimulation simultaneously to movements
Use prone exercises to stimulate righting reflexes that tend to imitate primitive motor function
Start seated and standing stimulation exercises to help decrease subluxation and modify synergy patterns
Attempt to increase passive range of motion (PROM) with gentle slow motion, rhythmic stabilization, or voluntary contraction followed by relaxation or gentle stretching.
Avoid vigorous traction on the arm when stretching connective tissue around the spastic joint
Use of electric stimulation can enhance muscle relaxation
Use the functional arm to simultaneously train the paretic arm to improve ROM and proprioceptive stimulation.
Use modalities (eg, ice, transcutaneous electrical nerve stimulation [TENS], vibration) to diminish spasticity
Constraint-induced movement therapy
Constraint-induced movement therapy (CIT) is a family of therapies that induce patients who have had a stroke to greatly increase the amount and quality of movement of their paretic limb, in turn improving function. CIT is based on the theory of "learned nonuse," first described by Wolf and colleagues  and later by Taub and coauthors. 
Substantial neurological injury leads to a shocklike phenomenon, called diaschisis, with dramatically depressed motor neuron function. During this shock period, the patient is unable to move the affected limb and subsequently learns to compensate with the functional limb. As the shock resolves and function starts to improve, attempts to use the affected limb result in clumsy and ineffective movements; according to the learned nonuse theory, this difficulty positively reinforces continued compensation.
Treatment begins by restraining the functional limb during all waking hours, except for specified activities, and then forcing the patient to perform tasks almost exclusively with their paretic limb for up to 2 weeks. This usually produces measurable improvement of function in the paretic limb, as well as increases in speed and strength of contraction, provided the patient has some selective hand movement (slight wrist and finger extension), good balance, and good cognitive and communication skills.
As reported by Morris, a behavioral training technique called shaping often is used in conjunction with CIT.  Shaping has resulted in substantial improvement of motor function.
Shaping approaches a desired motor outcome in small successive steps through explicit positively reinforced feedback by the therapist. This allows subjects to experience successful gains in performance with relatively small amounts of motor improvement.
A battery of approximately 60 tasks has been developed with a preliminary shaping plan for each task. Each task can be broken down into subtasks. Performance regressions are never punished and usually are ignored. If performance continues to exhibit no improvement after approximately 3 trials, the subject is encouraged to improve further at a later time, a simpler subtask is attempted, or an entirely different task is substituted. Eventually, an individualized task-oriented home program that emphasizes the use of the most impaired movements and joints is established.
Researchers report that patients tend to reach a plateau in motor recovery within 6-12 months following stroke. Taub and coauthors refuted this by studying the effectiveness of CIT in overcoming learned nonuse in chronic hemiplegic stroke patients. 
Compared with an attention-comparison group, the restrained subjects improved on each measure of motor function (ie, performance time, quality of movement, range of activities); in most cases, patients improved markedly. Two-year follow-up revealed that ADL functions had been maintained or increased. Researchers subsequently concluded that the use of CIT proved to be an effective means of restoring substantial motor function in chronic stroke patients.
In the past, surgical release of tendons and muscle was commonly performed on patients experiencing prolonged spasticity and synergy. For patients experiencing a painful spastic shoulder, surgical transection of the subscapularis and pectoralis tendons was performed to eliminate internal rotation and adduction forces.
Hecht reported that following treatment, up to 88% of these patients had improved pain and increased ROM, with some developing active abduction.  Today, this form of treatment rarely is used.
Without appropriate care, patients with hemiplegia have an increased risk of developing numerous shoulder complications, including nerve pressure palsies, nerve traction injuries, rotator cuff pathology, capsulitis, impingement syndromes, or subluxation. During the acute flaccid stage, care for the shoulder must take into account the position of the extremity in relation to gravity and body position. Techniques used in the prevention of poststroke shoulder pain and its complications must be utilized to neutralize undesirable or injurious positions.
Wheelchair armrests, lap trays, or forearm troughs are commonly used while the patient is in his or her wheelchair. Rigid armboards often are preferred to the use of slings while the patient is in the wheelchair because they allow the humeral head to approximate the glenoid fossa at a more natural angle and are less restrictive. Armrests also benefit the patient as the arm is in a nondependent position, thereby decreasing the incidence of edema. Armrests also can be used as an alternative for patients with decreased trunk control.
Even though optimal positioning is mandated, Kaplan suggested that judicious ROM exercises (through therapy) should be started within 24 hours poststroke. 
When moving patients in bed, or transferring them in and out of the wheelchair, positions of dependent arm traction should be avoided. When passive transfers are performed, the hemiplegic arm needs to be supported by holding the scapula rather than pulling on the arm.
About one third of stroke patients studied by Wanklyn and colleagues required assistance during transfers and tended to be at risk for incorrect handling by their caregivers, subsequently predisposing them to the development of shoulder pain.  Traction and excessive shoulder movement need to be avoided in order to prevent impingement or a rotator cuff tear.
When positioning the patient, caregivers should maintain the patient in reflex-inhibiting postures in order to avoid common hemiplegic complications, including spasticity and contractures. A literature review by Carr and Kenney found consensus on some positions considered proper for hemiplegic patients.  However, there was disagreement for other positions regarding the 9 "key points of control", as follows:
Head and neck
Many of the authors also warned that hemiplegic patients should avoid supine-lying positions as much as possible because abnormal reflex activity is highest in this position.
Since the scope of this article is specific to the shoulder, discussion of proper positioning is limited to the head and neck (which have an indirect effect on the upper extremity and the upper limb. The consensus position for the head and neck was found to be midline or turned to the affected side, which encourages patient attention to the environment of the affected side and may be beneficial for those patients with neglect. If sidebent, turn away from the affected side.
The shoulder should be protracted, with the arm brought forward to counteract the scapular tendency for retraction. The arm should be positioned in varying degrees of external rotation, abduction, and flexion. Positioning for the remainder of the upper extremity is as follows:
Elbow - Extension
Forearm - Supination
Wrist - Neutral
Fingers - Extended
Thumb - Abducted
The above positions are not supported by all authors, including Cailliet. Cailliet has recommended that the head be laterally flexed and rotated toward the unaffected side, and that the hand and fingers be supported in a wrist-extended and finger-flexed position. 
Carr and Kenney have stated that "current understanding seems to suggest that attendance to posture is likely to be an important element in maximizing patients' functional gains and quality of life."  For this reason, emphasize patient and caregiver education regarding proper positioning.
Slings often are used early poststroke in an attempt to prevent subluxation. Cailliet has contended that it continues to be the best method for supporting and protecting the hemiplegic shoulder while the patient is standing or transferring.  However, excessive sling use should be avoided due to the increased incidence of contractures.
Kirshblum has proposed that the following considerations be used when deciding on the use of a sling  :
Proper fit that promotes proper glenohumeral alignment (poor alignment can contribute to increased flexion synergy)
Protection of the flaccid extremity during transfers, standing, and ambulation (slings can interfere with balance, however)
Should not interfere with patient function
Should be relatively easy to don and doff
Should not create new problems (eg, edema in the dependent hand), contribute to synergy patterns, or cause scapulohumeral malalignment
Zorowitz and colleagues tested 4 different shoulder sling models for their efficacy in correcting subluxation in stroke patients and found that the only sling that significantly corrected vertical asymmetry was the single-strap hemisling in 55% of subjects, while total asymmetry was corrected most by the Rolyan support in 45% of subjects.  They contend that lateral displacement of the humeral head does not appear to be an inherent quality of subluxation, but the use of certain slings, especially the Bobath and Cavalier supports in this study, were found to contribute.
Brooke and coauthors found that the hemisling gave significantly better vertical correction compared with the Bobath sling, while the arm trough/lap board tended to overcorrect.  Their results also showed that the Bobath sling horizontally distracted the glenohumeral joint significantly more than the other 2 supports. Even though improved glenohumeral asymmetry was found in some cases, none of the slings consistently prevented subluxation in all cases.
Yu and colleagues also described the propensity of slings to contribute to the deleterious effects of joint immobilization and their promotion of undesirable synergy patterns. [22, 23] For this reason, no consensus has been reached amongst researchers or clinicians as to which model should be used to attain a particular therapeutic goal, or if they should be used at all.
Strapping also has been studied as a means for shoulder support. Theoretically, it should support the glenohumeral joint or reduce subluxation while allowing the upper extremity to move freely.
A study by Hanger and colleagues concluded that strapping the shoulder did not significantly preserve ROM or reduce the prevalence of subluxation over a 6-week trial, even when done concomitantly with standard physical therapy.  There was a trend toward improved pain and shoulder function, but it was not found to be statistically significant. These researchers also found that the presence of neglect or sensory loss at baseline was associated with poor outcome.
Other literature suggests that strapping has potential for reducing the incidence or the severity of hemiplegic shoulder pain, but those studies were small or uncontrolled. Hanger and coauthors also expressed that there are different strapping techniques that may be more effective than the one they used. 
Carr and Kenney reported that about two thirds of all stroke survivors will be disabled, up to 50% will be severely disabled, and 10-15% will require institutional care.  Motor weakness also is reported in 50-80% of survivors poststroke.
Brandstater reported that most spontaneous recovery of voluntary motor function occurs in the first 2-3 months following stroke. However, it can occur years later. 
Cailliet reported an unfavorable prognosis for complete upper extremity motor recovery if the flaccid stage lasts longer than 2 weeks,  while Carroll found this to be the case after only 1 week. 
Other unfavorable predictors in estimating functional recovery include excessive spasticity and impaired sensation and perception, with Van Buskirk and Webster reporting that sensory loss occurs in up to 80% of stroke patients. 
Depression also can contribute to unfavorable outcome, with Wanklyn and coauthors reporting a 22-27% incidence within the first few weeks poststroke. 
Kingery reported that the prognosis for resolution with preserved ROM is better in patients with some voluntary movements, with less spasticity, and without significant sensory loss.  Nearly 35% of patients with CRPS type 1 have symptom resolution in 1 year.
What would you like to print?
- Hemiplegic Shoulder Pain
- Shoulder Anatomy
- Pathophysiology of Shoulder Dysfunction
- Epidemiology of Shoulder Pathology
- Etiology of Poststroke Shoulder Pain
- Differential Diagnosis
- Laboratory Studies
- Bone Scans
- Additional Procedures
- Treatment of Poststroke Shoulder Pain
- Physical Therapy
- Surgical Intervention
- Patient Education
- Show All