Hemiplegic Shoulder Pain: Implications for Occupational Therapy Treatment

Gilmore, P., Spaulding, S., Vandervoort, A (2004). The Canadian Journal of Occupational Therapy, 71(1) 36-47.

Abstract
Hemiplegic shoulder pain is common after stroke causing hemiplegia. It adversely affects the recovery of arm function and independence in activities of daily living. Subluxation, abnormal tone and limited range of motion or capsular constrictions have been reported as potential causes. Other factors such as rotator cuff tears, brachial plexus injury, shoulder-hand syndrome and other pre-existing pathological conditions may also be associated with hemiplegic shoulder pain. The etiology remains unclear, but hemiplegic shoulder pain may result from a combination of the above factors. Scope. This literature review examines the possible causes of hemiplegic shoulder pain and discusses the implications for occupational therapy treatment. Occupational therapy interventions include proper positioning, facilitation of movement through purposeful therapeutic activities, increasing passive range of motion, implementation of external supports and treatment of shoulder-hand syndrome. Practice Implications. Understanding the processes involved will assist with effective assessment, treatment and prevention of hemiplegic shoulder pain. This will facilitate clients' participation in rehabilitation programs and move them towards attainment of optimal function.

Introduction
Cerebral vascular accident (CVA), or stroke, may cause a person to experience hemiplegia (paralysis) or hemiparesis (weakness) on the side of the body opposite the site of the CVA. Occupational therapists work with individuals who have experienced a CVA to help them optimize function in areas that are important to them. The development of hemiplegic shoulder pain (HSP) is a common clinical consequence of hemiplegia (Snels, Beckerman, Lankhorst & Bouter, 2000), affecting a person's ability to perform activities of daily living (ADL). An occupational therapist's goal is to optimize function in occupational performance tasks such as the client's ability to take part in activities of daily living (Trombly, 1989). The incidence of HSP in the literature varies, with reports ranging from 48% to as high as 84% (see Table 1). Shoulder pain usually develops in the second week after stroke, but could begin at an earlier or later stage (Chantraine, Baribeault, Uebelhart & Gremion, 1999). Wanklyn, Forester and Young (1996) found that 63.8% of 108 study participants developed HSP at some time during their 6-month study period. Thirty-nine reported HSP at discharge from hospital, 59 at 8 weeks post-discharge and 36 at 6 months. Thus, the number of people with HSP increased 8 weeks after discharge from hospital and by 6 months, HSP had resolved in some cases.

Shoulder function is imperative for optimal occupational performance, such as successful transfers, maintaining balance, performing ADL and for effective hand function (Rizk, Christopher, Pinals, Salazar & Higgins, 1984). Roy, Sands, Hill, Harrison and Marshall (1995) found strong positive associations between HSP and prolonged length of stay in hospital, arm weakness, poor recovery of arm function and poor ADL function in a study of 76 people following a first stroke. There were also indications of lower rates of discharge to home. The development of HSP can interfere with and prolong rehabilitation (Griffin, 1986). A painful shoulder may also limit a client's general mobility. When clients protect their arms because of pain, they reduce both active and passive movements (Andersen, 1985). Restricting movement may limit the client's ability to perform activities of daily life including rolling in bed, transferring, putting on a shirt and donning/doffing socks and shoes (Andersen, 1985; Davies, 2000). Occupational therapists play a role in the rehabilitation of people with HSP attempting to optimize their occupational performance. Several authors have been interested in the possible causes and treatment for HSP (Andersen, 1985; Griffin, 1986; Roy, 1988; Teasell & Heitzner, 1998; Van Langenberghe, Partridge, Edwards & Mee, 1988;). The purpose of this paper is to review the literature for the etiology and treatment of HSP and report the implications for occupational therapy.

Definition of hemiplegic shoulder pain
There appears to be lack of a widely accepted definition or set of diagnostic criteria for HSP. "Hemiplegic shoulder pain remains a nebulous clinical entity, defined differently by each investigator" (Griffin, 1986, p. 1884). The diagnosis of HSP is a clinical one, based on signs and symptoms (Griffin, 1986). Pain and decreased passive range of motion (ROM) at the shoulder are often included in the criteria for HSP.

Griffin (1986) indicates that pain is an elusive symptom, defying quantitative measurement even in clients with intact cognitive, perceptual and communication skills. As a result, the criterion for pain does not seem to be clearly defined or standardized in the literature (Van Langenberge et al., 1988). The client's experience of shoulder pain has been variously defined as the presence or absence of pain at rest and during active and passive movement and/or palpation (Kumar, Metter, Mehta & Chew, 1990; Poulin de Courval, Barsauskas, Berenbaum, Dehaut, Dussault, Fontaine, et al. 1990). Pain has been rated from mild to severe as well as present all the time, at night or with movement (Wanklyn et al., 1996). In one study, pain was not quantified at all, but subjectively evaluated by therapists who compared clinical responses (i.e. facial expressions and vocalizations) to shoulder ROM (Hecht, 1992). Methods of quantifying pain have been described in the literature. For example, investigators have used vertical visual analogue scales to quantify pain (Dekker, Wagenaar, Lankhorst, & de Jong, 1997; Ikai, Tei, Yoshida, Miyano & Yonemoto, 1998; Roy, Sands & Hill, 1994; Zorowitz, Hughes, Idank, Ikai & Johnston, 1996).

As indicated earlier, reduced amplitude of passive shoulder ROM is often included in the indicators of HSP (Griffin, 1986; Rizk et al., 1984). However, differences exist in the investigators' descriptions of ROM. Passive ROM has been considered to be limited or not limited (Roy et al., 1994; Wanklyn et al., 1996) or meeting criteria for capsulitis (Roy et al., 1994). Some investigators report goniometric measurements for shoulder ROM (Ikai et al., 1998; Joynt, 1992; Kumar et al., 1990). Investigators also use different types of goniometers such as a fluid filled goniometer to measure shoulder ROM (Dekker et al., 1997).

As noted in the 1980's, and still relevant is that a common definition for HSP needs to be developed so that information can be collected in a systematic way (Van Langenberghe et al., 1988) and findings between studies can be compared for incidence and treatment effectiveness (Griffin, 1986).

Anatomy of the shoulder joint
The shoulder is a ball and socket joint referred to as the glenohumeral joint (Martin, 1985). This joint involves the bony structures of the glenoid cavity, and scapula including the acromion and head of the humerus (Wood, 1989). The humerus fits into the glenoid cavity, which is shallow and tends to be unsupportive. The ligaments that enclose the joint are loose allowing for a large degree of motion. The shoulder joint also lacks support for bones and ligaments and depends on four surrounding muscles to maintain its integrity. These four muscles are called the rotator cuff and fuse with the joint capsule to strengthen the joint (Martin, 1985). The long head of the biceps brachii muscle also helps in holding the head of the humerus in the glenoid cavity (Martin, 1985). Biomechanically, stability in the shoulder joint has been sacrificed in favour of mobility (Davies, 2000). It is not surprising that the shoulder becomes a focus of pain (Van Langenberge et al., 1988) because this joint has a large range of movement, making it unstable and vulnerable to dislocation (Wood, 1989). It is important for occupational therapists to understand the complex anatomy of the shoulder joint in order to give appropriate care to clients with HSP.

Factors that contribute to hemiplegic shoulder pain
HSP may develop two weeks after stroke, but can develop earlier (Chantraine et al., 1999) or after several months (Davies, 2000). HSP must not be accepted as a symptom of a CVA, because shoulder pain is not present at the onset of a stroke that causes hemiplegia. Therefore, something must have happened subsequent to the CVA to cause the shoulder pain (Davies, 2000). The issue of what causes HSP is complex and there are a number of alleged factors that may contribute. Some of the factors that contribute to HSP are subluxation, abnormal muscle tone and limitations to shoulder ROM, such as capsular contractures or adhesive capsulitis. Other factors such as rotator cuff tears, brachial plexus injury, shoulder-hand syndrome and the influence of pre-existing pathological conditions may also be associated with HSR Shoulder pain may develop as a result of a combination of these factors (Andersen, 1985; Griffin, 1986; Van Langenberghe et al., 1988). However, there is disagreement in the literature, therefore the etiology continues to remain unclear.

Subluxation
Shoulder subluxation is defined as changes in the mechanical integrity of the glenohumeral joint causing a palpable gap between the acromion (upper part of the scapula) and the head of the humerus (Teasell & Heiztner, 1998). The inferior subluxation of the head of the humerus is a common problem in hemiplegia and has been reported to have an incidence of up to 80% (Faghri, Rodgers, Glaser, Bors, Ho & Akuthota, 1994). During the initial period following stroke, the hemiplegic arm is flaccid (Cailliet, 1991; Teasell & Heitzner, 1998). In the flaccid stage there is a total loss of deep tendon reflexes and loss of muscle tone of the affected side. The client is not able to initiate active motor function (Cailliet, 1991). Therefore, the rotator cuff muscles cannot perform their function of maintaining the head of the humerus in the glenoid cavity and there is a risk of shoulder subluxation (Teasell & Heitzner, 1998).

Other factors also seem to play a role in subluxation of the shoulder joint. Basmajian and Bazant (1959) (as cited in Teasell & Heitzner (1998)) proposed that in the "normal" state of the shoulder, subluxation was prevented by the upward angulation of the glenoid cavity. It was hypothesized that after a stroke causing hemiplegia, the upward angulation of the scapula would be lost, leading to subluxation. Cailliet (1991) also stated that in the hypotonic (flaccid) stage after stroke the scapula depresses and downwardly rotates therefore predisposing the head of the humerus to undergo inferior subluxation. However, Culham, Noce and Bagg (1995) studied 34 subjects with hemiplegia to determine if scapular and humeral orientation differed between the affected and non-affected shoulders in 2 groups of subjects (low muscle tone on the affected side and high muscle tone on the affected side) to determine if there was a relationship between these groups and shoulder subluxation. Findings indicated little evidence of a consistent pattern of alteration in shoulder joint orientation, particularly in subjects with increased muscle tone. There was no support for the concept of a relationship between scapular and humeral orientation and shoulder subluxation. This study concluded that scapular position is not an important factor in subluxation (Culham et al., 1995).

Controversy continues to exist about whether or not subluxation of the shoulder joint causes shoulder pain in hemiplegia (see Table 2). Many studies have shown that a possible cause of shoulder pain is subluxation (Chantraine et al., 1999; Poulin de Courval et al., 1990; Roy et al., 1994 & Van Ouwenaller, Laplace & Chantraine, 1986). The incidence of subjects with shoulder pain and subluxation has been found to be between 26% and 100%. As well, Roy et al. (1994) found that 90% of the patients with subluxation experienced more severe pain at rest and on movement then did patients without subluxation. Poulin de Courval et al. (1990) found that subjects with shoulder pain had more severe subluxation of the affected shoulder than those without shoulder pain. However, Bohannon and Andrews (1990), Zorowitz et al. (1996) and Ikai et al. (1998) found no association between subluxation and shoulder pain. Therefore, not all people with subluxed hemiplegic shoulders experience shoulder pain. Findings in these studies may differ because of sample sizes, differences in measures used to assess pain and confounders such as subjects with preexisting shoulder injuries that may or may not cause shoulder pain (i.e. rotator cuff tears).

Shoulder subluxation occurs frequently in association with the initial flaccid stage after a hemiplegic stroke. The flaccid musculature of the shoulder and not scapular rotation appear to account for shoulder subluxation (Teasell & Heitzner, 1998). Shoulder subluxation may be a cause of HSP, however this has not been definitively established and some authors suggest otherwise (Teasell & Heitzner, 1998).

Abnormal muscle tone
Cailliet (1991) described 4 stages of muscle tonicity experienced by a person who has had a CVA including the transischemic attack stage, the flaccid stage, the spastic stage and the synergy stage. Two of these stages, the flaccid stage and the spastic stage, may directly relate to HSP. A flaccid upper extremity of the affected side may cause shoulder subluxation, which as previously discussed, may be a factor in HSP. In addition, the affected limb(s) may progress to the spastic stage, which is an increase of muscle tone to the affected side (Cailliet, 1991). Teasell and Heitzner (1998) describe spasticity as " a disorder of motor function characterized by a velocity-sensitive increase in resistance to passive stretch of muscles accompanied by hyperactive tendon reflexes and often associated with a clasp-knife phenomenon" (p. 492). The anatomy of the shoulder joint makes it vulnerable when the muscle tone is altered because of the lack of the normal voluntary and reflex muscle activity (Davies, 2000). Davies (2000) reported that in hemiplegia, if a person experiences pain and loss of ROM at the shoulder, it is likely that the shoulder mechanisms have been disturbed by abnormal and unbalanced muscle tone.

Some clinical observations suggest that pain does not occur until the spastic stage (Davies, 2000) and there is some experimental research to support this. Van Ouwenaller et al. (1986) looked at factors in 219 subjects followed for one-year post stroke. The researchers identified a much higher incidence of shoulder pain in people with spastic (85%) rather than flaccid (18%) hemiplegia. Another study supported this finding by demonstrating that the subjects with shoulder pain had significantly more spasticity of the affected limb than those subjects without pain (Poulin de Courval et al., 1990).

In the hemiplegic shoulder, the normal, coordinated and timed movements of the scapula and humerus known as "scapulo-humeral rhythm" have been disturbed by abnormal and unbalanced muscle tone (Nasser Bierman & Atchinson, 2000). More specifically, flexor tone can predominate in the hemiplegic upper extremity and may result in scapular retraction and depression as well as internal rotation and adduction of the shoulder (Teasell & Heitzner, 1998). This is called the "synergy pattern" and is inevitable where recovery of hemiplegia is incomplete. This pattern causes spastic shoulder muscle imbalance (Teasell & Heitzner, 1998). Agonist muscles in the synergy pattern tend to shorten. Shortened muscles inhibit movement, reduce ROM and prevent other active movements (Teasell & Heitzner, 1998). The subscapularis is a shoulder muscle that assists in internal rotation, abduction and flexion of the shoulder (Martin, 1985). In normal movement, external rotation of the shoulder is required for arm abduction over 90 degrees (Teasell & Heitzner, 1998). To create this movement the subscapularis muscle relaxes to allow external rotation to occur. In the flexor synergy pattern, subscapularis muscle tone is increased, limiting shoulder external rotation, abduction and flexion (Teasell & Heitzner, 1998). Hecht (1992) showed statistically significant improvements in shoulder ROM and a subjective reduction of shoulder pain following phenol blocks to the nerves to the subscapularis muscle in clients with hemiplegia. This suggests that increased tone of the subscapularis muscle is the chief cause of shoulder pain in clients with spastic hemiplegia when external rotation is limited (Hecht, 1992). Therefore, spacticity may interfere with the normal scapulo-humeral rhythm during movement of the shoulder and increases the probability of shoulder joint contracture and shoulder pain (Van Ouwenaller et al., 1986; Poulin de Courval et al., 1990).

In contrast, Joynt (1992) studied 67 patients with shoulder problems following hemiplegia and concluded that pain was unrelated to spasticity, strength or sensation. Bohannon, Larkin, Smith and Horton (1986) completed a statistical analysis of 50 consecutive patients with hemiplegia of whom 36 had shoulder pain. These authors also found that spasticity was unrelated to shoulder pain. Studies have found contradicting evidence that spasticity is a factor in HSP (see Table 3). However, growing evidence appears to exist for spasticity being the cause of HSP. This is convincing given the large sample sizes of the studies that found a relationship between spacticity and shoulder pain (Van Ouwenaller et al., 1986 & Poulin de Courval et al., 1990) and the subjective decrease in shoulder pain noted in 100% of the subjects given phenol blocks to the nerves of the spastic subscapularis muscle in another study (Hecht, 1992).

Limitations to shoulder ROM, capsular contractures or adhesive capsulitis
Hemiplegia has been associated with the development of shoulder capsular contractures or adhesive capsulitis. Adhesive capsulitis has been identified as a source of pain in the spastic hemiplegic shoulder (Bohannon et al., 1986). Inability to move the hemiplegic arm, normally, may lead to capsular limitations. Attempts to move in the direction opposite to the contracture may elicit pain due to stretching (Van Langenberghe et al., 1988). Zorowitz et al. (1996) found in a study of 20 subjects that shoulder pain was strongly correlated with the limitation of external rotation at the shoulder joint. Shoulder pain was not correlated to age, vertical, horizontal or total asymmetry, shoulder flexion, abduction or Fugl-Meyer Motor Assessment scores (Zorowitz et al., 1996). Capsular restriction was also noted to occur in 30% of hemiplegic shoulders in a random sample of stroke patients with HSP (Hakuno, Hironobu, Ohkawa, & Itoh, 1984). Rizk et al. (1984) performed shoulder arthography on 30 people with a painful hemiplegic shoulder and found that 23 (77%) had demonstrated capsular constriction typical of frozen shoulder. Ikai et al. (1998) studied 74 people with hemiplegia and also found that there were correlations between shoulder pain and limited shoulder ROM especially external rotation. Adhesive changes were noted on arthograms indicating that adhesive capsulitis may be the main cause of shoulder pain. Furthermore, studies by Joynt (1992) and Bohannon et al. (1986) showed shoulder pain was related most to loss of shoulder motion. The finding by Bohannon et al. (1986) suggested that shoulder pain demonstrated by people with hemiplegia might be a manifestation of adhesive capsulitis. Researchers have reported that adhesive capsulitis causing limited shoulder ROM played a more important role in shoulder pain than subluxation (Ikai et al., 1998). Interestingly, Teasell and Heitzner (1998) state,

The presence of a frozen or contracted shoulder likely occurs as a consequence of the spastic muscle weakness or imbalance ..., although its presence may be simply an association with no etiologic significance to pain. Alternatively, contracted shoulder muscles may be the final mechanism by which spastic muscle imbalance leads to shoulder pain (p.494). The studies reviewed appear to support decreased shoulder ROM and adhesive capsulitis as main factors in HSR

Other potential causes of hemiplegic shoulder pain
Other potential causes of HSP include rotator cuff tears, brachial plexus injury, shoulder-hand syndrome and the effects of pre-existing pathological conditions.

Shoulder pain is common among people who have not had a stroke and this pain is often associated with rotator cuff tears. Therefore, it is not surprising that rotator cuff tears have been seen as potentially associated with HSP in clients who have had a stroke (Teasell & Heitzner, 1998). In the study conducted by Rizk et al. (1984) mentioned earlier in this paper, no evidence of rotator cuff tears were seen on arthrography taken of 30 patients reporting HSP. As well, Hakuno et al. (1984) also using arthrography, found no significant difference in the frequency of rotator cuff tears in the affected and unaffected shoulders of 77 patients with hemiplegia. These researchers found that rotator cuff tears correlate with shoulder pain before stroke, but not after the onset of hemiplegia. Joynt (1992) studied 67 clients with hemiplegic shoulder pain. Some of the patients (28) received an injection of 1% lidocaine to see if there would be a marked decrease in pain. Of these patients, 50% obtained moderate or marked relief of pain and improved ROM. This study indirectly implicates rotator cuff tears as the cause of HSP. Even in clients that are asymptomatic, partial rotator cuff tears are common and it is difficult to diagnose if the tears were present premorbidly (Teasell & Heitzner, 1998). Therefore, only weak evidence exists to support the association of rotator cuff tears and HSP.

Brachial plexus and peripheral nerve injury have been reported as a potential complication in the client with hemiplegia after stroke leading to shoulder pain and subluxation (Griffin, 1986). Brachial plexus injury has often been related to flaccidity, however little is reported about the causal relationships with hemiplegia and pain (Van Langenberghe et al., 1988). Positive identification of the brachial plexus injury is also difficult because it is often masked by the symptoms of a CVA, such as sensory deficits (Meredith, Taft & Kaplan, 1981). Cailliet (1991) argues that traction applied to the arm in the flaccid state may lead to plexus lesions possibly resulting in HSP. Chino (1981) conducted electromyographic investigations on 21 people who had a stroke and concluded that initial flaccidity may lead to overstretching of the brachial plexus, leading to peripheral nerve injuries and contributing to the development of subluxation which was discussed previously as a possible cause of HSP. Kingery, Date and Bocobo (1993) completed physical examination, needle electromyography and nerve conduction studies across the brachial plexus within 4 months after stroke on 50 people with hemiplegia. These researchers were unable to make a diagnosis of brachial plexus injury or proximal mononeuropathy in any participant. Brachial plexus injuries have been attributed primarily to improper handling and positioning of the affected side (Meredith et al., 1981). Therefore, some evidence exists supporting subluxation with superimposed traction resulting in lesions of the person with flaccid musculature (Van Langenberge et al., 1988).

Shoulder-hand syndrome, also known as reflex sympathetic dystrophy or reflex dystrophy syndrome (Davies, 2000) has been defined as sympathetically mediated pain which may involve the hemiplegic arm (Teasell & Heitzner, 1998). Shoulder-hand syndrome is notable in the upper extremity by pain and edema, hyperesthesia, protective immobility, trophic skin changes and vasomotor instability (Teasell & Heitzner, 1998; Griffin, 1986). Initially, shoulder pain develops, followed by limited ROM and an edematous, painful hand and wrist (Teasell & Heitzner, 1998; Griffin, 1986). In shoulder-hand syndrome passive flexion at the wrist and finger joints are painful and ROM is often limited because of swelling over the back of the hand (Teasell & Heitzner, 1998). As the condition progresses, pain settles and stiffness predominates. The skin becomes shiny and cyanotic (Griffin, 1986). The hand may atrophy if untreated and pain may resolve suddenly after several weeks (Teasell & Heitzner, 1998). This syndrome may not be uncommon in hemiplegia. Van Ouwenaller et al. (1986) found the incidence of shoulder-hand syndrome to be 28% in a study of 219 people with hemiplegia. In the same study, of the people with HSP (157/219), 23% had signs and symptoms of shoulder-hand syndrome. The researchers of this study also found a higher incidence of shoulder-hand syndrome in people with spasticity (27%) than in subjects with flaccidity (7%).

Shoulder-hand syndrome usually presents within 3 months of onset of stroke and rarely presents later than 5 months (Davis, Petrillo, Eichberg & Chu, 1977). This extremely painful condition may interfere with a client's overall rehabilitation and if untreated, may lead to permanent deformity of the hand and fingers preventing future functional use of the upper extremity (Davies, 2000). Few studies have been completed exploring shoulder-hand syndrome as a potential cause of HSP and the present research does not definitively correlate shoulder-hand syndrome with HSP.

Pre-existing pathological conditions may exist prior to a stroke causing hemiplegia and may contribute to HSP. In normal aging, shoulder flexion and abduction tends to become painful (Griffin, 1986), therefore limiting ROM at the shoulder. One reason for this decreased ROM is postural changes. People who are elderly may develop increased dorsal kyphosis causing downward scapular rotation (Cailliet, 1991). This posture causes impingement early in shoulder movement by lowering the coracoacrominal arch (Cailliet, 1991). Other reasons contributing to decreased shoulder ROM and pain are the degenerative changes in articular surfaces and soft tissue that occur with age. With normal use, wear and tear cause degenerative changes of the rotator cuff and biceps brachii tendon. This musculature becomes edematous and the adjacent bursa often thicken (Griffin, 1986). The result may be reduced space under the coracoacromial arch of the shoulder with impingement as the likely cause of pain (Griffin, 1986). Clinicians need to be aware that these degenerative changes may or may not have been asymptomatic before the onset of hemiplegia (Griffin, 1986) and may contribute to shoulder pain.

Implications for occupational therapy treatment
HSP usually develops in a typical pattern (Davies, 2000). Occupational therapists need to be aware of this pattern to effectively treat clients with hemiplegia. The client often starts to report sharp pain at the end range of movement at the shoulder, when the arm is being moved during therapy. Clients can usually identify the painful location. If potential causal factors are not addressed, over time the pain increases rapidly and the client may describe pain on all movement, especially with flexion and abduction of the shoulder. The client may also experience pain only in certain positions or at night time when lying in bed. Severe sudden pain may occur, not only at full range of motion, but also when lowering the arm or during certain stages of movement. As time progresses, the client may find it increasingly difficult to report the exact location of the pain. If therapeutic intervention does not relieve discomfort, the client tends to report pain day and night and cannot tolerate any arm movement. Intolerable pain will then impede rehabilitation efforts. Occupational therapy treatment of HSP may consist of prevention including: positioning and handling techniques, facilitating active movement, increasing passive range of motion, implementing external supports such as slings, and treating shoulder-hand syndrome (see Fig. 1).

Positioning and handling
Davies (2000) believes that "when the predisposing causes of the painful shoulder are carefully avoided, the condition can be prevented altogether" (p. 345). She also reiterates, "prevention and successful treatment of the painful shoulder are dependent upon all members of the team understanding and avoiding the possible causes" (Davies, 2000, p. 346). Occupational therapists are often part of the team that is involved in a client's care throughout the healthcare process, including; acute care, inpatient rehabilitation, outpatient therapy (i.e. Day Hospital) and community intervention. Therefore, the occupational therapist should advocate that team members attend to the client's position when lying in bed, sitting in a chair and when assisted to move and transfer. As well, it is important that occupational therapists educate the client, and family members in proper positioning and handling techniques because everyone who assists the client has the responsibility of caring for the hemiplegic shoulder (Andersen, 1985; Griffin, 1986).

Carr and Kennedy (1992) reviewed the positioning of a person who has had a CVA. These authors reported broad agreement in the literature that the affected upper extremity should be placed in various positions including the shoulder protracted, arm brought forward, spine aligned and fingers extended (Carr & Kennedy, 1992). The arm should also be supported when at rest (Andersen, 1985) (i.e. with pillows). However, controversy appears to surround the degree of positioning of the arm in the forward position while in sitting and lying (Carr & Kennedy, 1992). Andersen (1985) reported that the client's scapula must be protracted, with the shoulder flexed more than 90 degrees and the elbow extended when positioned on the affected side. When positioned on the unaffected side, the hemiplegic arm should also be positioned with the scapula protracted, the shoulder flexed more than 90 degrees, the elbow extended and the arm supported on pillows (Andersen, 1985). These positions place the affected arm in a reflex-inhibiting situation preventing or limiting spasticity and preventing potential shoulder pain (Andersen, 1985).

The side-lying positions are essential for scapular movement (Davies, 2000). Wood (1989) reported that in the early stages of hemiplegia, a client should not be cared for while in supine (which is typically done) as this encourages retraction of the scapula, protraction of the shoulder girdle and forward arm extension. When sitting, clients should be encouraged to position their arm on a table with the shoulder girdle protracted and their elbow on the table for proper support (Andersen, 1985). While sitting in bed, clients should have their hemiplegic arm supported by pillows (Wood, 1989). If a client is sitting in a wheelchair, the therapist may recommend a lapboard to position the arm (Andersen, 1985). When turning or transferring a client, it is essential that the affected arm be supported and the trunk brought forward by grasping behind the scapula and not pulling on the affected arm (Andersen, 1985; Wood, 1989). Pulling on the affected arm can distort the shoulder causing trauma or subluxation (Wood, 1989).

Occupational therapy is a client-centred profession, believing in the worth of the individual (Canadian Association of Occupational Therapists (CAOT), 1991). Occupational therapists believe that successful treatment will only be achieved if there is active participation from the client in the rehabilitation process. Therapists work together with clients to set goals and decide on treatment activities that will assist the client with optimal functional independence in the future. Therefore, the client must be educated to inform the therapist when any movement or position is painful. This feedback is the only method that therapists have to avoid damaging structures. It is usually impossible to feel or see the moment of injury (Davies, 2000).

Facilitation of active movement
Frequently, occupational therapists are involved in remediation of the affected hemiplegic arm to optimize a client's ability to participate in daily occupations. When the clients perform active ROM activities, they should avoid the combined motion of scapular retraction with forward flexion of the arm because this can cause impingement due to improper scapulo-humeral rhythm (Andersen, 1985). Davies (2000) reports that practicing active arm elevation too vigorously can cause painful trauma. If a client who has inadequate scapular control is encouraged to practice lifting the arm actively, the failure of the scapula to provide a stable origin for the working muscles will result in impingement and pain (Davies, 2000). The occupational therapist and client should work together to choose activities that encourage normal scapulo-humeral rhythm of the shoulder. To be beneficial, these activities need to be meaningful and purposeful for the client. It is important to note that during activities requiring the client to reach for or place objects, the position of these materials in front of the client at or below waist level encourages and assists scapular protraction and forward flexion of the shoulder (Andersen, 1985). The occupational therapist must also guide the movement to ensure scapular protraction, upward rotation and shoulder flexion. Some activities include reaching for and grasping a glass of water or attempting to dust a tabletop. The height of the activity can be gradually raised as the client improves (Andersen, 1985) encouraging further ROM.

Increasing passive range of motion
Maintaining passive ROM of the shoulder joint is an essential prophylactic measure to prevent joint stiffness and soft tissue contractures (Griffin, 1986). The loss of external rotation at the shoulder appears to be a cause of shoulder pain in clients with hemiplegia (Ikai et al., 1998; Zorowitz et al., 1996). Kumar et al. (1990) analyzed 28 people with hemiplegia who were randomly assigned to one of three exercise groups during treatment. Pain developed in 8% of the people in the ROM by therapist group, 12% of the people in the skate board group developed pain and 62% of the people in the overhead pulley group developed pain. The three groups did not differ significantly in terms of side of involvement, extent of hemiplegia or presence of subluxation. The authors concluded that the development of pain seems to depend on the type of exercise received regardless of the presence or absence of shoulder subluxation. The group using overhead pulleys had the highest risk of developing shoulder pain. The incidence of pain was low for the people who received passive shoulder ROM by a therapist. The authors recommended that the use of overhead pulleys should be avoided at least early in rehabilitation of people following a stroke.

How much ROM is enough? Therapists seem to take each joint through its full ROM, however, full ROM for one person may be different than for another (Griffin, 1986). Perhaps a more realistic expectation is the maintenance of pain free functional ROM at the shoulder. Functional ROM at the shoulder has been defined as flexion to 100 degrees, abduction to 90 degrees, external rotation to 30 degrees and internal rotation to 70 degrees (Griffin, 1986). Davies (2000) also reports that to begin passive ROM the client must sufficiently trust the therapist not to cause pain. If that trust is not established between therapist and client, anticipation of pain prior to ROM exercises may increase muscle tone, and defeat the purpose of the activity.

Apprehensive clients may be asked to complete self-assisted ROM exercises moving the arm as far as they can without causing pain, using the unaffected arm for power, with the hands clasped together. In this position, external rotation will be assured and clients gain control of their therapy by knowing they can stop the movement at any time. Clients must learn to move their own shoulder correctly, in order to prevent injury (Davies, 2000). Goal-oriented, purposeful activities such as washing one's own face with clasped hands ensures protraction of the scapula and may help a client move without fear of pain because the ROM maintained while concentrating on the activity (Davies, 2000).

External supports
The use of external support systems, including slings and lapboards, are controversial. External supports are implemented to realign the shoulder joint in an effort to reduce subluxation and to protect the arm from trauma (Cutler Lewis, 1989). The use of slings have both advantages and disadvantages. One advantage is that slings remain the best method of supporting a flaccid hemiplegic arm when a client is transferring, standing and ambulating (Teasell & Heitzner, 1998). Slings are often used in the initial flaccid stage after a CVA (Teasell & Heitzner, 1998). However, slings encourage flexor synergies, inhibit arm swing during ambulation, contribute to contracture formation and may decrease a client's body image and encourage the client to avoid use of the affected arm (Teasell & Heitzner, 1998). As well, "reduction of subluxation to prevent pain has not been fully substantiated" (Spaulding, 1999, p. 174). However, until research determines that there are no long-term consequences of pain in relation to subluxation, attempting to reduce shoulder subluxation appears to be an appropriate treatment (Spaulding, 1999). Authors supporting this approach recommend therapists attempt to reduce subluxation, whatever the cause (Van Ouwenaller et al., 1986; Roy et al., 1995). If such a support is appropriate for an individual, it is the role of the occupational therapist to provide clients with an external support in order to position the shoulder and prevent or correct misalignment (Trombly, 1989).

However, there is discontentment among occupational therapists concerning the most beneficial method to support the hemiplegic arm (Spaulding, 1999). Numerous external devices have been used and evaluated to support the subluxed shoulder (Boyd & Gaylard, 1986; Zorowitz, Idank, Ikai, Hughes & Johnston, 1995; Boyd, Pepin & Szabo-Hartin, 1999 & Spaulding, 1999). Boyd et al. (1999) investigated the use of shoulder supports by therapists with stroke survivors in Canada and compared the results of two survey questionnaires carried out in 1984 and 1994. For a description of the supports, see Boyd et al. (1999) research article. The use of a lapboard attached to a wheelchair, cuff type sling and arm trough support on the wheelchair were the most frequently used supports by therapists during both time periods. There was a significant decrease in the use of the Bobath axial roll over the ten-year period. Therapists surveyed in both 1984 and 1994 used the lapboard to increase a client's awareness of his/her arm, the cuff type sling to reduce gait problems and the arm trough to protect the affected arm (Boyd et al., 1999). However, the comparison of the two surveys indicated "growing skepticism by therapists about the effectiveness of using these supports" (Boyd et al., 1999, p. 167) and the need for evidence to inform this practice.

Some evidence for the use of external supports exists. Zorowitz et al. (1995) measured the ability of four ambulatory shoulder supports to correct shoulder subluxation. The four supports tested were the single-strap hemi-sling, the Bobath roll, the Roylan humeral cuff sling and the Cavalier support. The single-strap hemi-sling assessed corrected vertical asymmetry to a significant degree (Zorowitz et al., 1995). These findings were consistent with Spaulding's (1999) biomechanical analysis of four support systems including; two shoulder slings, the Bobath axial roll and a laptray attached to a wheelchair. Results indicated that slings with straps over the unaffected shoulder gave continuous support to a flaccid hemiplegic arm and that the Bobath axial roll may cause unwanted lateral force. As well, Spaulding (1999) found that lapboards should be maintained at an appropriate distance from a hemiparetic shoulder to be beneficial. This research contributes to occupational therapy's body of knowledge by understanding the consequences of external support systems (Spaulding, 1999). It has also been noted that occupational therapists must have knowledge of muscle tone, potential for pain, range of motion requirements, perceptual limitations and level of independence in occupational performance when choosing to use and selecting a support system that will be beneficial for an individual client (Spaulding, 1999).

Treatment of shoulder-hand syndrome
Occupational therapists must be aware of the possible link between shoulder-hand syndrome and HSP. Occupational therapy objectives for shoulder-hand syndrome are prevention, early recognition and treatment initiation (Andersen, 1985). These objectives comprise a portion of the treatments available for shoulder-hand syndrome. Prevention of shoulder-hand syndrome incorporates many of the treatments previously discussed including early ROM exercises to counteract the effects of immobilization (i.e. edema and contractures of the shoulder and finger joints), active movements of the affected limb and proper positioning of the hemiplegic arm during ADL and rest. As well, Davies (2000) advocates the use of a wrist cock-up splint to avoid wrist flexion, reduce edema and pain and for protection of the vulnerable hand. However, splinting is somewhat controversial as splinting the wrist and/or hand immobilizes these joints. Clients with shoulder-hand syndrome may also be hypersensitive and therapists can be instrumental in providing desensitization programs (Teasell & Heiztner, 1998). A client's emotional state is often associated with this syndrome (Griffin, 1986). The client may appear tense, overemotional or emotionally dependent (Griffin, 1986). Occupational therapists must also consider psychological management, including teaching relaxation techniques to control pain and anxiety.

Other treatments for HSP that have been shown to reduce shoulder pain in the literature include corticosteriod injections (Dekker et al., 1997; Joynt, 1992; Snels et al., 2000), functional electrical stimulation (Faghri et al., 1994; Chantraine et al., 1999) and motor blocks (Hecht, 1992). These treatments have less direct implications for occupational therapy treatment, however these management options are beneficial for therapists to be aware of when working with this population.

Summary
Hemiplegia is common after stroke and, as a consequence, HSP may develop (Teasell & Heitzner, 1998). The etiology of HSP is unclear. HSP may be a result of a combination of factors. However, it appears that the majority of problems begin with abnormal muscle tone and immobilization of the shoulder, which is characteristic of hemiplegia of the upper extremity. Controversy exists as to whether or not subluxation is the cause of shoulder pain in hemiplegia, however limited external rotation of the shoulder caused by immobilization or spastic subscapularis muscle function appears to correlate with HSP (Teasell & Heitzner, 1998). It is important for occupational therapists to understand the basic processes that may contribute to HSP so that they can more effectively assess and treat these clients (Andersen, 1985). Health care professionals including occupational therapists must consider HSP to be an important problem to address (Snels et al., 2000). To prevent or alleviate shoulder pain, therapists should direct efforts towards promotion of proper positioning and handling of the affected upper extremity and activities that promote proper ROM to avoid impingement of the joint structures (Andersen, 1985). The occupational therapist and client need to work together to choose purposeful therapeutic activities that encourage scapulo-humeral rhythm of the shoulder. Furthermore, careful prescription of external devices, such as slings or lapboards should be considered in some cases. The absence of pain will facilitate clients' participation in rehabilitation programs moving them towards attainment of optimal function in daily life.