Classification scheme for medical complications of stroke

Prospective studies suggest that direct effects of ischemic stroke, such as hemorrhagic conversion or cerebral edema, account for the majority of deaths within the first week, but medical complications account for at least 50% of mortality thereafter (154; 65).

Medical complications of stroke decrease survival up to several years after discharge (06). Admission to stroke units significantly reduce the complications of stroke, including immobility and mortality within the first 4 weeks (04). Prevention of aspiration by speech and swallow evaluations, avoidance of urinary catheters, and prophylactic anticoagulation against venous thromboembolism are beneficial. Later, the most frequent and severe complications are venous thromboembolism, pneumonia, urosepsis, cardiac arrhythmias, and myocardial infarction. Other complications include falls, pressure sores, psychiatric disorders, and central post-stroke pain (154).

The classification scheme for post-stroke medical complications, which includes neurologic sequelae (recurrent stroke, post-stroke seizures), is provided in Table 1 (91). Several of these topics (sleep apnea, vascular dementia, depression, dysphagia, normal pressure hydrocephalus) are discussed in other clinical summaries. The topics discussed in this article are listed in Table 2.

Table 1. Medical Complications of Stroke: A Classification Scheme

Table 2. Medical Complications of Stroke

Deep venous thrombosis and pulmonary embolism

Epidemiology. Venous thromboembolism is a major cause of morbidity and mortality after acute ischemic stroke. In the absence of prophylaxis, approximately 50% of patients develop deep venous thrombosis within 2 weeks after a stroke, mostly below the knee. Deep venous thrombosis may begin on the second day after a stroke and reach a peak incidence within 7 days (16). The risk of deep venous thrombosis correlates with the degree of paralysis (90). Before prophylaxis with anticoagulation became routine, clinically apparent deep venous thrombosis was associated with mortality from pulmonary embolism in 37% of cases (20). Pulmonary embolism is responsible for about 5% to 10% of deaths in the acute period and 25% after the acute period (70). In a series of patients with ischemic and hemorrhagic stroke, pulmonary embolism was not diagnosed in half of the patients before death (183). Sudden death occurred in 50% of patients with pulmonary embolism.

Among patients with ischemic stroke discharged between 1979 to 2003, 1.17% had venous thromboembolism, 0.74% had deep venous thrombosis, and 0.51% had pulmonary embolism; in those with hemorrhagic stroke, the rates were higher, 1.93%, 1.37%, and 0.68%, respectively (155).

Venous thromboembolism etiology is multifactorial and involves Virchow's triad: vessel wall damage, venous stasis, and hypercoagulability (146; 75). Damage to the vessel wall prevents the endothelium from inhibiting coagulation and initiating local fibrinolysis.

The risk factors for arterial stroke and venous thromboembolism overlap significantly. Obesity, cigarette smoking, and hypertension are independent risk factors for pulmonary embolism, whereas hypercholesterolemia and diabetes mellitus are not (49). Venous stasis due to immobilization or obstruction leads to accumulation of activated coagulation factors. Following stroke, the paretic leg is preferentially affected due to a repeated minor trauma and loss of the calf muscle pump (79). Additionally, atrial fibrillation, oral contraceptives, hormone-replacement therapy, and several inherited thrombophilias (such as hyperhomocysteinemia and activated protein C resistance) predispose to both venous thromboembolism and ischemic strokes (120; 48).

Other risk factors include severity of the illness, age, dehydration, delayed institution of preventive measures, hemorrhagic stroke, cryptogenic ischemic stroke, previous venous thromboembolism, morbid obesity, and prothrombotic conditions, including contraceptive use (144; 08).

Pulmonary embolism can be asymptomatic or massive, leading to sudden death. Almost 40% of asymptomatic patients with deep venous thrombosis have evidence of pulmonary embolism on lung scans (115). Although 90% of emboli originate in the lower extremities, ultrasonography of the leg veins was positive in only 29% of patients with pulmonary embolism (168). The remainder emboli arose mostly from pelvic deep venous thrombosis (70).

Differential diagnosis. Edema caused venous stagnation in the paretic limb may mimic deep venous thrombosis. Respiratory failure may be due to cerebral herniation or a brainstem lesion. The post-thrombotic syndrome (PTS) attributed to venous hypertension and abnormal microcirculation (15) is characterized by edema and pain with or without venous ulceration (79).

Diagnostic workup. The clinical suspicion should guide the choice of further testing. Most patients with pulmonary embolism have at least 1 of these symptoms: sudden onset dyspnea, chest pain, syncope, and hemoptysis (110).

All patients should be assessed for pretest clinical probability of pulmonary embolism (Tables 3 and 4). Three scoring systems were developed: Wells (176), Geneva (178), and Pisa (111). Only after the clinical probability was determined should further studies be performed.

D-dimer level is elevated in patients with various systemic illnesses, like myocardial infarction, pneumonia, heart failure, cancer, recent surgery, and stroke (48). A negative test result combined with low pretest probability of disease by clinical criteria excludes thrombotic disease without the need to perform ultrasound.

In patients with high pretest probability, the result should not affect clinical decision (177). Venous duplex ultrasonography is highly sensitive for deep venous thrombosis, especially in symptomatic patients (95). Rarely, if there is a high clinical concern for deep venous thrombosis with a normal ultrasound, venography may be considered (77). In many patients with pulmonary embolism, the venous thrombus is not detectable.

Conventional pulmonary angiography is the gold standard but is invasive and is not widely available. Spiral CT of the chest is sensitive and may identify small distal pulmonary emboli with a potential for growth (48). Echocardiography may detect right ventricular failure due to pulmonary embolus (48).

In an unselected cohort with acute stroke, MRI detected venous thromboembolism in 40%, proximal deep venous thrombosis in 18%, and pulmonary embolism in 12% of patients (78).

In elderly patients with typical vascular risk factors, evaluation for systemic coagulopathy is low yield (19). Immobilization is the main risk factor for deep venous thrombosis in this population. However, approximately 20% of patients younger than 45 years of age may have a hematologic disorder (84).

In patients with high clinical probability, a D-dimer level below cutoff does not exclude the diagnosis of venous thromboembolism. This needs to be confirmed with lower limbs ultrasonography or helical CT scan of the chest (145). However, CT alone may not be sensitive enough to exclude pulmonary embolism in these patients (141).

Table 3. Wells Prediction Rule for Diagnosing Deep Venous Thrombosis

Table 4. Wells Prediction Rule for Diagnosing Pulmonary Embolism

Prevention of thromboembolism. Nonpharmacologic prevention of deep venous thrombosis includes early mobilization, graduated compression stockings, and pneumatic sequential compression devices.

The simplest preventive measure is early mobilization (03). Bed rest following stroke should be restricted to patients with large cerebellar or cerebral lesion and increased intracranial pressure, severe angina and myocardial infarction, deep venous thrombosis (prior to anticoagulation), and postural ischemia due to large-vessel carotid or vertebrobasilar artery disease. If possible, avoid intravenous lines in the paretic arm as they may increase the risk of deep venous thrombosis.

The thigh-length graduated compression stockings (GCS) did not reduce the incidence of proximal deep venous thrombosis after stroke and caused skin complications (30).

However, intermittent compression stockings prevent deep venous thrombosis and increase survival in patients with acute stroke (29). The American College of Chest Physicians recommends the use of intermittent pneumatic compression devices or elastic stockings for patients who have contraindications to anticoagulants (76).

Unfractionated heparin 5000 IU sc given 3 times daily reduced venous thromboembolism following acute stroke from 73% to 22%, along with a decrease in the combined rate for deep venous thrombosis, pulmonary embolism, and death (106).

Low molecular weight heparins may be given once daily and have a lower risk for thrombocytopenia and osteoporosis than unfractionated heparin. A meta-analysis showed that low molecular weight heparin had the best benefit-to-risk ratio in patients with acute stroke (69; 153). The once-daily administration is also cost effective (138).

For acute stroke patients, either low molecular weight heparin, low-dose unfractionated heparin, or danaparoid may be used (76). Heparin may be used after 24 hours from thrombolysis; the combination with aspirin may be safely used later.

In an observational study, the addition of sequential compression devices to unfractionated heparin reduced the risk of deep vein thrombosis more than 40-fold (70).

In patients with intracerebral hemorrhage, the benefits of antithrombotic medication need to be balanced with the risk of hemorrhage expansion. In a small, randomized study, initiation of low-dose unfractionated heparin (5000 U sc 3 times daily) on the second day from ictus significantly lowered the incidence of pulmonary embolism, compared with delayed (day 4 or day 10) therapy (11). In another prospective randomized study of 75 patients, low-dose enoxaparin (40 mg sc daily) initiated 2 days after onset was as effective and safe as the compression stockings (123). However, a Cochrane review found 2 randomized controlled trials totaling 120 patients that failed to provide evidence for or against the use of anticoagulants for venous thromboembolism prophylaxis (134). A randomized controlled trial, stopped prematurely because of low recruitment, and a meta-analysis that included these patients showed that anticoagulation for venous thromboembolism prevention is safe but did not significantly reduce the incidence of venous thromboembolism, pulmonary embolism, or death (127).

In patients at risk for heparin-induced thrombocytopenia (HIT), fondaparinux, an indirect factor Xa inhibitor, 2.5 mg sc is a safe alternative (55).

The new oral anticoagulation (NOAC) agents have not yet been tested in patients with ischemic stroke. Nevertheless, 4 systematic reviews, network meta-analyses (NMAs), and cost-effectiveness analyses of randomized controlled trials have not shown strong evidence that they should replace postoperative low molecular weight heparin in primary prevention of venous thromboembolism. For acute treatment and secondary prevention of venous thromboembolism, there is little evidence that new oral anticoagulants are more efficacious than warfarin. However, the risk of hemorrhagic complications was lower for some new oral anticoagulants than for warfarin (159).

Treatment of venous thromboembolism in stroke patients. Initiation of treatment in patients with suspected venous thromboembolism depends on the strength of the clinical suspicion, how rapidly the results of the confirmatory tests are available, and the location of the deep venous thrombosis or pulmonary embolism.

If the clinical suspicion is high, parenteral anticoagulant is indicated while waiting for the results of diagnostic tests. If the suspicion is intermediate, initiation of parenteral anticoagulation is indicated if the results of diagnostic tests are likely to be delayed for more than 4 hours. In patients with low suspicion, there is no need to initiate parenteral anticoagulation provided the results will be available within 24 hours.

If deep venous thrombosis is isolated to a distal segment of the leg, and there are no severe symptoms or risk for extension, serial imaging for 2 weeks is reasonable. However, if the distal deep venous thrombosis is accompanied by severe symptoms and there is risk of extension, initiation of anticoagulation over serial imaging is recommended. Extension of the distal thrombus on serial imaging warrants anticoagulation. In patients with venous thromboembolism, once-daily low molecular weight heparin or fondaparinux is preferred over unfractionated heparin.

Some patients prefer to avoid the post-thrombotic syndrome (PTS) and accept a higher risk of bleeding. They may benefit from catheter-directed or systemic thrombolysis.

If anticoagulation is contraindicated, like in intracerebral hemorrhage, a temporary inferior vena cava (IVC) filter is recommended. Anticoagulation should be resumed if the contraindication no longer exists. However, insertion of IVC filter is not recommended in addition to anticoagulation.

The duration of therapy for the first unprovoked venous thromboembolism is 3 months if the risk of bleeding is high and extended if the risk of bleeding is low or moderate. At the end of 3 months, the risk-benefit ratio should be reassessed before further recommendations are made. If venous thromboembolism is provoked by either surgery or a transient nonsurgical factor, the duration of anticoagulation is 3 months. If cancer is present, extended anticoagulation is recommended regardless of the bleeding risk.

Compression stockings for 2 years or more are indicated in patients with symptomatic deep venous thrombosis of the leg or post-thrombotic syndrome. If they fail to relieve the symptoms, a trial of intermittent compression device is reasonable.

If acute pulmonary embolism causes hypotension (systolic BP < 90 mmHg) or the patient is at risk for developing hypotension, intravenous thrombolytic therapy through a peripheral vein is indicated provided there is no increased risk for hemorrhage. Analysis of a National Inpatient Sample (NIS) from 2009 to 2012 revealed that systemic thrombolysis is associated with increased risk of hemorrhagic stroke compared with catheter-directed intervention, although mortality was similar (98). In this situation, catheter assisted thrombectomy is appropriate. In critical situations in which thrombolysis or catheter assisted thrombectomy failed or the patient is in shock and likely to die within a few hours, surgical pulmonary thrombectomy may be beneficial (76).

The standard treatment for venous thromboembolism is intravenous unfractionated heparin and oral warfarin initiated on the same day. Unfractionated heparin is continued for a minimum of 5 days or at least 24 hours of INR 2 or above. The ideal INR target is 2 to 3. Warfarin dosage should begin with 5 mg, not higher (57; 33). Outpatient anticoagulation clinics are preferred in terms of anticoagulation control, patient outcomes, and health care costs (26).

If warfarin is contraindicated, such as during pregnancy, unfractionated heparin or a low molecular weight heparin, such as enoxaparin, may be used. Early mobilization and stopping procoagulant agents (eg, hormone-replacement therapy or oral contraceptives) are both advisable.

In patients with cancer, low molecular weight heparin is preferred to warfarin. Additionally, if the patient is unreliable, long-term subcutaneous injections of a low molecular weight heparin is an alternative.

The American College of Chest Physicians recommends low molecular weight heparin or vitamin K antagonists to the newer oral agents (76). More information regarding the efficacy of the newer anticoagulants is needed.

Patients with acute venous thromboembolism taking dabigatran, a direct thrombin inhibitor, 150 mg po bid were compared with warfarin after intravenous heparin. Dabigatran was non-inferior and was associated with fewer bleedings than warfarin (151).

Rivaroxaban was also proven effective for deep venous thrombosis and pulmonary embolism treatment. A dose of 15 mg po twice daily for 3 weeks followed by 20 mg daily was noninferior to enoxaparin followed by warfarin (40; 41).

Moreover, apixaban was not inferior to enoxaparin followed by warfarin, yet caused fewer clinically relevant bleeding. The dose of apixaban was 10 mg po bid for 7 days, followed by 5 mg po bid for 6 months (01).

Bone disease: osteoporosis and fractures

Stroke causes limited mobility and instability, which is associated with osteoporosis and leads to an increased risk of falls and fractures. Most fractures occur late after stroke onset on the hemiplegic side (144). Hip fractures are associated with an increased risk of institutionalization and death (43).

Epidemiology. Out of 130,000 discharged acute stroke patients, 2.0% suffered fractures by 1 year and 10.6% by 10 years (37). The risk of hip fracture increases 4-fold after stroke (71).

Pathophysiology. Bone strength is a composite of bone density and bone quality, with the latter thought to include bone architecture, bone damage (eg, microfractures), and mineralization (147). Immobilization of the limbs after stroke leads to osteoporosis (144). After stroke, 30% of fractures occur in the upper extremities (129).

Clinical risk factors that contribute to fracture risk independent of bone mineral density are presented in Table 5 (72).

Table 5. Risk Factors for Osteoporotic Fractures in Women

Bone loss may result from disuse following hemiparesis. However, a study found only weak associations with motor function, activities of daily living, or ambulation (144).

Vitamin D deficiency increases the risk of hip fracture in elderly patients with disabling stroke symptoms (147). Hypovitaminosis D results from immobilization, advanced age, malnutrition, and sunlight deprivation (150). Medications such as warfarin, anticonvulsants, and proton pump inhibitors can also contribute to this deficiency (148; 150; 100). Dabigatran, a direct oral thrombin inhibitor, is associated with a lower risk of osteoporotic fractures compared to warfarin (92).

Trauma leading to fractures is more likely to occur in stroke patients whose symptoms include weakness, numbness, neglect, imbalance, or decreased awareness (139).

Clinical features. Vertebral fractures are associated with chronic back pain, spinal deformity, functional limitations, and increased risk of hospitalization and mortality (68; 44).

Osteoporosis is defined as bone mineral density of 2.5 SD or more below the average value for premenopausal women (T score, less than -2.5 SD). In the presence of 1 or more fragility fractures, osteoporosis is considered severe (72).

Table 6. World Health Organization (WHO) Definition of Osteoporosis

The goals of osteoporosis treatment are prevention of fractures, increased bone mass, treatment of fracture and skeletal deformity, and improvement of physical function (121). A pyramidal approach to treatment has been recommended (122). The base of the pyramid consists of lifestyle changes, including adequate calcium and vitamin D intake, physical activity, and fall prevention. Physical activity is needed for bone formation and maintenance, but walking alone does not reduce fracture risk. Exercises that reduce the risk of falling by improving mobility, muscle function, and balance could reduce fracture risk (45). To reverse osteoporosis after hemiplegia, daily weight training for minimum 60 and 90 minutes was needed for males and females, respectively (56).

Calcium and vitamin D reduce the risk of fracture (36). The recommended dietary intake of calcium is 1000 mg/day for men and women aged 50 years or younger and 1200 mg/day for those older than 50 years of age (NIH Dietary Supplement Fact Sheet--Calcium). The recommended dose of vitamin D is 400 IU/day for men and women aged 51 to 70 years and 600 IU/day for those 71 years or older (NIH Dietary Supplement Fact Sheet—Vitamin D). The second level includes addressing and treating secondary causes of osteoporosis. The third level includes pharmacotherapeutic interventions to improve bone density and reduce the risk of fracture.

Bisphosphonates inhibit bone resorption. Etidronate disodium did not significantly reduce the fractures (149). Other bisphosphonates include alendronate, ibandronate, and risedronate.

Other drugs available are raloxifene, teriparatide, calcitonin, and certain estrogens. Raloxifene and strontium ranelate may increase the risk of venous thromboembolism and fatal stroke and should be avoided in patients at risk for stroke (173).

Complications of stroke

Falls. Traumatic brain injury is one of the most devastating and feared complications of falls, making up 78% of fall-related deaths and 79% of the cost (160). Stroke doubles the risk of falling (66).

Balance problems while dressing was the strongest risk factor for falls after stroke, whereas residual motor symptoms are not associated with increased risk of falling (89). In another study, most falls occurred during transfers to and from a wheelchair or bed (133).

Depression following stroke and TIA has been associated with accidental falls (161). Independent predictors of depression and anxiety were female sex, younger age, and higher socioeconomic deprivation score (17). Impaired executive-controlled processes, frequent in stroke patients, negatively affect the balance and mobility and contribute to falls (103). These factors compounded by comorbidities, polypharmacy, decreased vision, and often decreased cognition in stroke patients pose a greater risk of falling.

Detailed cognitive testing and formal gait and balance testing are used by stroke rehabilitation centers. Screening tests such as the Postural Assessment Scale for Stroke patients (PASS) and the Postural Control and Balance for Stroke test (PCBS) are valuable clinical measurements of postural control in stroke patients (07; 140) and can play a role in assessing the risk of fall.

Fall prevention programs provide education about supervision of high-risk patients, proper seating and wheelchair transfers, and regular toileting.

Pressure sores. Pressure sores occur in many bedridden stroke patients (22). They can develop and progress quickly, and if infected, they can cause great morbidity and mortality.

A Swedish retrospective study of 161 patients with stroke recorded 116 pressure ulcers, 30 patients having more than 1 ulcer (58). Most often, the pressure sores occur in the lower body, with the sacrum being most frequently affected at 22.1% (53).

Limited mobility causes sustained pressure and ischemia of the skin over the weight-bearing points, usually the bony prominences. Other risk factors include diabetes, peripheral vascular disease, urinary incontinence, and low body mass index (10).

Pressure sores are classified according to stages (118).

Stage I is nonblanchable erythema on intact skin.

Stage II is a partial thickness skin loss of the dermis presenting as a shallow open ulcer with a red or pink wound bed, without slough.

Stage III is a full thickness tissue loss involving the subcutaneous tissue or fascia. Bone, tendon, and muscle are not exposed.

Stage IV is a full thickness tissue loss with exposed bone, tendon, or muscle.

The Braden Scale predicts the risk of developing pressure sores. The scores range from 6 to 23, with higher risk associated with lower scores (09; 14).

Bed sores can be prevented by turning every 2 hours, pressure-reducing devices, good skin care (especially in those with fecal incontinence), daily checks, and high protein nutritional intake (166).

Pressure sore treatment consists of using proper wound dressing and removal of the necrotic debris to prevent bacterial growth and infection (51).

Peripheral nerve injury. Compression peripheral nerve injury in stroke patients results from limb malposition and may cause additional disability or pain. The risk of injury increases with sensory loss, weakness, neglect, and limb edema. In addition, use of manual wheelchairs increase the likelihood of an upper limb nerve injury like the median and ulnar nerves (12). Hematoma, a complication of anticoagulation, may result in brachial plexus injury or femoral nerve entrapment (42).

Electrodiagnostic studies help establish the nerve injury site and assess the severity of injury and the need for surgical intervention. In most cases, splinting and other supportive devices as well as pain management may be sufficient.

Pain syndromes

Many patients with stroke suffer from pain, which impairs quality of life. A prospective study of 443 patients with stroke showed that 29.56% suffered from pain (14.06% in acute, 42.73% in the subacute, and 31.90% in the chronic stroke stage). Headache manifested acutely, musculoskeletal and central post-stroke pain was seen more often in the subacute and chronic stage, and spasticity-related pain was prevalent in the chronic stage (130).

Central post-stroke pain. This pain is caused by a lesion in the central nervous system. Central pain is described as a burning, aching, throbbing, cramp-like sensation or a combination of different types of pain and is often provoked by a non-noxious stimulus. Patients older than 70 years of age tend to experience non-burning pain. Characteristic to central pain is the sensory loss to pinprick or temperature in the same region; however, the spinothalamic sensation may still be preserved in many patients (87).

In a systematic review of 69 papers, central post-stroke pain was reported in 11% of strokes in any locations and in 50% of patients with medulla and thalamus strokes. Central post-stroke pain coincided with acute stroke in 26% of patients (97).

Lesions of the ventral posterolateral, ventral medial, and medial dorsal nucleus and the trigemino- and spinothalamocortical pathways, including the brainstem and cerebral cortex, lead to post-stroke pain (13; 32; 119). A rare location for pure central post-stroke pain is the posterolateral region of the mesencephalon involving the spinothalamic tracts. Because of paucity of associated symptoms, it may be mistaken for malingering or psychogenic pain (31).

The medial lemniscus pathway has a modulatory role. One woman with well controlled central post-stroke pain due to lateral medullary stroke redeveloped central post-stroke pain after a second infarct involving the ipsilateral medial medullary region (81). Lesions of the ventral caudal nucleus suggest involvement of other pathways in central pain development (112; 80).

The type of pain does not correlate with the site of the lesion (13). Pain sites can range from the entire half of the body to a restricted limb or part of a limb. Imaging of injury to the spinothalamic strokes may be achieved with diffusion tensor tractography (61).

Central post-stroke pain is a diagnosis of exclusion, and other causes for pain must be considered. Central post-stroke pain can be resistant to standard use of analgesics and opioids. There is lack of high-quality clinical trials.

Amitriptyline has been effective in a small, randomized controlled study (94); however, there is a need for more data (113).

Pregabalin is effective at escalating doses of 150, 300, and 600 mg/day (174). In addition, it may improve pain-related anxiety and sleep disturbances (83).

Carbamazepine is probably effective for chronic neuropathic pain, but there is a lack of information beyond 4 weeks (181). The effect of gabapentin on neuropathic pain is probably not superior to carbamazepine (179).

Lamotrigine in dose of 200 mg daily was moderately effective in a small randomized, double-blind, placebo-controlled, cross-over study (172). However, a Cochrane review provided no convincing evidence of benefit at doses of 200 to 400 mg/day. In addition, its titration is difficult because of risk for rash (180). Levetiracetam is not effective for post-stroke pain treatment (67; 182).

Intravenous lidocaine and morphine improve central pain; however, their delivery method is limited to short-term use and carries a large side-effect profile (05). Intrathecal baclofen may improve central post-stroke pain (162). Fluvoxamine and mexiletine may also be used as adjuvants for pain treatment (82).

A potentially useful noninvasive method is repetitive transcranial magnetic stimulation—rTMS (96). Fiber tracking with diffusion tensor imaging may predict response to rTMS (50). However, there is a significant publication bias (38). A systematic review of Transcranial Direct Current Stimulation (tDCS) on neuropathic pain in patients with stroke shows promise, but there is high heterogeneity (35; 143).

There is evidence that graded motor imagery (like limb laterality recognition, imagined movements, and mirror movements) and mirror therapy provide clinically meaningful improvements in pain and function (114; 165). For multimodal physiotherapy, electrotherapy, and manual lymphatic drainage, the evidence is absent or of very low quality (156).

Botulinum toxin, BTX-A, has been used with some success for neuropathic pain; however, a clinical trial is needed to confirm the initial results (60).

An interesting effect of cold caloric stimulation is rapid relief of thalamic pain (142; 107; 108). Refractory thalamic pain was successfully treated by stellate ganglion block (99). In 1 case report, ultrasound guided block of the stellate ganglion with lidocaine had lasting effect at 9 months (102).

Motor cortex stimulation is the preferred neurosurgical method to control intractable central pain. The motor prefrontal cortex seems to be the optimal site of stimulation because the procedure is less invasive, achieves better pain control (48% vs. 25% for deep brain stimulation vs. 7% for spinal cord stimulation), and is less likely to trigger painful sensations (74). For nonthalamic lesion, spinal cord stimulation achieved relief for more than 12 months in 44.4% of patients (164). Motor cortex stimulation is also useful for painful brainstem and spinal cord lesions (163).

Pain relief from motor cortex stimulation can persist at 12-month follow-up (158; 171). The long-term results were excellent (over 80% reduction in pain) in 31% of patients, satisfactory (50% to 80% reduction) in 23% of patients, and no improvement in 15% of patients (157). Epidural hematoma and subdural effusion are potential complications of this procedure.

The mechanism of action of motor cortex stimulation is thought to be increasing of regional cerebral blood flow to the ipsilateral corticothalamic connections (18) or anti-inflammatory effects of stimulation (152).

Other useful locations for electrical stimulation include peri-ventricular grey matter (117; 137) and the centromedian thalamic nucleus (02).

Unfortunately, the data on neurosurgical management of central post-stroke pain are limited, and further studies are needed to confirm these findings (34; 46).

Shoulder pain. Shoulder pain was reported by 22% of patients within the first 4 months of the first stroke. Seventy-nine percent of these patients had moderate to severe pain. Loss of arm motor function and a high NIHSS score correlate with development of shoulder pain. This symptom limits daily functioning and the rehabilitation process (101). A retrospective study of 786 patients with stroke shows a decrease of shoulder pain frequency over the last 15 years, suggesting an improvement of both stroke and rehabilitation therapies (109).

Shoulder pain after stroke may be caused by shoulder subluxation, proximal arm spasticity, local traumatic injury, shoulder-hand syndrome, and complex regional pain syndrome. The severity of paralysis and its stage (flaccid or spastic) alter the anatomy of the shoulder joint differently and determine the type of shoulder pain (169). It is unclear why pain resolves spontaneously in some patients but not in others. A study of 16 patients with shoulder pain revealed decreased adaptation to pain compared to a group of 14 patients with similar functional status but without shoulder pain (73).

Shoulder subluxation. Shoulder subluxation may result from the inferior displacement of the paretic arm (169). Glenohumeral subluxation was present in nearly 50% of patients in a case-control study of 107 hemiplegic adults with stroke within 30 days (126).

Subluxation can be detected by palpation, plain radiographs of the shoulder, or ultrasonography (131). Positioning and special attention to the hemiparetic limb is important in preventing subluxation, especially in those dependent on transfers for mobility (175). Multiple support slings have been used to reduce shoulder subluxation with mixed results (125). One study showed that the level of subluxation decreased in patients who did not use a sling and that wearing of an arm sling may impair this correction (170). Observational studies suggest that shoulder orthoses reduce vertical subluxation and pain (116).

Electrical stimulation of the shoulder successfully prevented and treated subluxation by maintaining muscle bulk and tone and by enhancing functional recovery through cortical feedback. Long-term follow-up was limited, however (169). A small, randomized pilot study failed to demonstrate additional benefit of kinesio-tape or neuromuscular electric stimulation (NEMS) to conventional therapy alone, which consisted of careful shoulder handling and daily mobilization (59).

Intramuscular electrical stimulation in 61 chronic stroke survivors with shoulder pain and subluxation significantly reduced pain levels. The effect was maintained at 12-month follow-up (23). However, a small single-blinded randomized trial showed better pain control with EMG-triggered neuromuscular electric stimulation compared to transcutaneous electrical nerve stimulation (TENS), both immediately and at 1-month follow-up (28). In a small case series of 5 selected patients, a fully implantable peripheral nerve stimulator was found to be safe and significantly reduced shoulder pain at 12-month follow-up (184).

A systematic review of neuromuscular electric stimulation showed reduction of shoulder subluxation in the acute and subacute phase but not in the chronic stroke or reduction of shoulder pain (93).

Spasticity. Spasticity of the shoulder following stroke is often painful and limits the range of motion. There is adduction and medial rotation of the arm associated with flexion at the elbow, wrist, and fingers. Different treatments have been employed, including medication for pain and spasticity, treatment of capsulitis, surgical release with transaction of the subscapularis tendon, and local botulinum toxin A injections.

Botulinum toxin type A injection in the pectoralis major reduced pain caused by spasticity in the first week after stroke in a double-blind, randomized clinical trial of 31 patients (105). Subscapularis muscle is another useful site for injection (185). However, botulinum toxin is not indicated for nonspastic causes of shoulder pain (86). Moreover, a meta-analysis of 950 patients showed that overall effectiveness of botulinum toxin type A does not seem to differ from placebo for post-stroke patients with upper limb spasticity (62).

Other promising therapeutic modalities include repetitive transcranial magnetic stimulation (27), suprascapular nerve block (136), modified wheelchair arm support (128), and extracorporeal shock wave therapy (63).

Frozen shoulder. Frozen shoulder, or adhesive capsulitis, caused by recurrent local injuries, is characterized by shoulder pain and limited motion in all directions. It is diagnosed by contrast enhanced arthrograms in 50% of hemiplegic patients with shoulder pain. In addition, rotator cuff tear was seen in 22% of patients (104).

Shoulder-hand syndrome. Shoulder-hand syndrome is a form of complex regional pain syndrome in the arm after stroke. It is characterized by severe shoulder pain, metacarpophalangeal joint tenderness, hand edema, changes in skin color and temperature, dysautonomia with excessive sweating, and hyperesthesia. Atrophy of skin and muscles of the shoulder and hand with sparring of the elbow often follows.

The mechanism is probably central and peripheral sensitization due to stroke (135), but involvement of the sympathetic system has not been demonstrated. Furthermore, shoulder subluxation and peripheral nerve damage increase the likelihood of developing complex regional pain syndrome (47).

The diagnosis of complex regional pain syndrome is primarily clinical. Diagnostic studies are used to exclude other processes. Bone scintigraphy may show increased periarticular uptake, particularly at the shoulder and wrist (52), and decreased bone mineral density in the paretic limb when compared to matched healthy controls (88).

Preventative measures consist of avoidance of shoulder trauma. Treatment is a combination of physical therapy and pain management. Mirror therapy may reduce pain and improves the function of the upper limb (21).

Those with severe sympathetic dysfunction may benefit from regional block, although strong evidence is lacking. Investigations into neuromodulation through spinal cord stimulation and administration of intrathecal analgesia have been undertaken in patients with complex regional pain syndrome (88).

Two systematic reviews of acupuncture found several heavily biased studies, suggesting the need for further randomized clinical studies (25; 132).

Heterotopic ossification. Heterotopic ossification causes of post-stroke pain and functional limitations and deposition of bony tissue in the soft tissue of the paretic limb (124). The calcification may also occur in the nonparetic limb (85; 54).

Urinary incontinence. Urinary incontinence within the first 7 days of stroke was noted in 53% of patients. One third of these patients remained incontinent at 12-month follow-up. Furthermore, those who were incontinent in the acute phase were 4 times more likely to be institutionalized after 1 year and a predictor of death (64).

Post-stroke urinary incontinence can be described based on etiology (160):

(1) Urge urinary incontinence: urgency followed by involuntary leakage that can result from a lesion of the central micturition pathways.

(2) Functional urinary incontinence: inability to achieve self-toileting due to impaired mobility from stroke.

(3) Stress urinary incontinence: involuntary leakage on effort such as coughing. This is usually present prior to stroke onset but is exacerbated by coughing associated with dysphagia and aspiration.

There are insufficient data from clinical trials to guide continence care in patients with stroke (167). Management of urinary incontinence after stroke is like that of the general population. Treatment begins with behavioral interventions like timed voiding, prompted voiding, bladder retraining with urge suppression, and pelvic floor muscle retraining and compensatory rehabilitation approaches (39). Pharmacological treatment and urologic consultation with surgical intervention may be necessary.