Introduction

First described more than 100 years ago, Complex Regional Pain Syndrome (CRPS) still remains a medical challenge today, with a natural history characterized by chronicity and relapses that can result in significant disability over time.1 Women are affected more frequently than men, but the overall incidence is unknown. Fractures and surgical insult are often the precipitating events, but CRPS can also develop after a seemingly benign trauma. Adding to this confusion was the myriad of names used to describe the syndrome, such as “Reflex Sympathetic Dystrophy”, “causalgia”, “Sudeck’s atrophy”, “algodystrophy”, “neurodystrophy”, and “post-traumatic dystrophy”. To standardize the taxonomy, the term CRPS was adopted in 1995 by the International Association for the Study of Pain (IASP).2 This revision was deemed necessary because previous names, such as Reflex Sympathetic Dystrophy, were misleading. The underlying pathophysiology is poorly understood, and patients often do not respond to sympathetic blockade. Diagnostic criteria were also proposed by the IASP.2 In contrast, treatment of CRPS remains a controversial topic. While the literature is replete with reports advocating the use of various clinical treatments for CRPS (from physiotherapy to spinal cord stimulation), the levels of supportive evidence are quite variable and sometimes limited. Accordingly, a literature search for level 1 evidence (from randomized controlled trials) was undertaken to determine the benefits associated with these therapeutic modalities. For the purpose of this review, no distinction was made between CRPS type 1 (formerly Reflex Sympathetic Dystrophy) and 2 (formerly causalgia).

Methods

Search strategy and selection criteria

The literature search for this review was conducted during the second week of April 2009 using the MEDLINE (January 1950 to the 2nd week of April 2009) and EMBASE (January 1980 to the 15th week of 2009) databases.

The following MeSH terms were searched: “Complex Regional Pain Syndrome”, “Reflex Sympathetic Dystrophy”, and “causalgia” as well as the key words “algodystrophy”, “Sudeck’s atrophy”, “shoulder hand syndrome”, “neurodystrophy”, “neuroalgodystrophy”, “reflex neuromuscular dystrophy”, and “post-traumatic dystrophy”. Results were limited to studies conducted on human subjects, written in English, and published in peer-reviewed journals. Only randomized controlled trials (RCTs) pertaining to the treatment of CRPS were considered for analysis. We excluded trials that investigated the impact of interventions on parameters measured in vitro (skin resistance, temperature, vasodilation, thresholds of pain to different stimuli, mapping of allodynic area) but did not assess the clinical response of patients. Furthermore, RCTs needed to deal exclusively with CRPS to be retained for analysis. We also excluded trials that enrolled a heterogeneous mix of patients suffering from CRPS and neuropathic pain (diabetic neuropathy, post herpetic neuralgia, phantom limb pain, trigeminal neuralgia, nerve root injury/radiculopathy, peripheral nerve injury/lesion/neuroma) or postsurgical pain (sustained after thoracotomy, mastectomy, inguinal hernia repair) and indiscriminately pooled results from all subjects. However studies that provided data specific to CRPS were included in this review. Randomized controlled trials (RCTs) published in the form of abstracts or correspondence were also excluded. After selecting the initial articles, we examined the reference lists as well as our personal files for additional material. No RCTs were excluded based on factors such as definition of intervention allocation or primary and secondary (clinical) outcomes. However, non-randomized studies, observational case reports, and cohort studies were excluded to avoid potential biases introduced by institutional practices.

Results

Our search yielded 50 RCTs pertaining to the treatment of CRPS. Seven of these were excluded because they investigated treatments requiring consultants (acupuncturists, qi gong masters) or modalities (electromagnetic field, transcranial magnetic stimulation, occlusive splinting, and hyperbaric oxygen therapy) that are not readily available to most practitioners. One identified trial was excluded as it had recently been retracted;Footnote 1 another RCT was excluded because both groups received the same treatment (graded in vivo exposure) (Appendix 1). Eighteen of the remaining 41 RCTs studied pharmacological treatment (Table 1) and 18 investigated intravenous regional blockade or central and peripheral nerve blocks (Table 2). Spinal cord stimulation and adjuvant therapy were addressed in one and four studies, respectively (Tables 3 and 4). Overall, the quality of the 41 RCTs was variable. The average enrolment was 31.7 subjects per study. The IASP definition of CRPS was used in 30.0% of trials. Blinded assessment and sample size justification were provided in 70.7% and 19.5% of RCTs, respectively. The duration of CRPS prior to enrolment was provided in 70.7% of studies and varied from 50 days to 13 yr. Pain was the most common endpoint studied (78.1% of trials); patient follow up varied from two weeks to five years.

Table 1 Randomized controlled trials pertaining to pharmacological treatment
Table 2 Randomized controlled trials pertaining to intravenous regional blockade and nerve blocks
Table 3 Randomized controlled trials pertaining to spinal cord stimulation
Table 4 Randomized controlled trials pertaining to adjuvant therapy

Pharmacological therapy

Calcitonin

Calcitonin has received considerable interest in the management of CRPS because of its analgesic properties through release of ß-endorphin as well as its inhibition of bone resorption.3 To date, four RCTs have investigated the use of calcitonin in CRPS.

Bickerstaff et al. 3 randomized 38 patients with upper extremity CRPS to a four-week regimen of nasal calcitonin or placebo. The authors observed no differences between the two groups in terms of pain, hand volume, stiffness, grip strength, and vascular/sudomotor changes. Furthermore, radiographic, densitometric, and scintigraphic evaluations of the metacarpal bones were also similar. Gobelet et al. 4 randomized 24 patients suffering from hand or foot CRPS to physiotherapy alone (eight-week course) or the same physiotherapy regimen combined with a three-week course of subcutaneous calcitonin. At eight weeks, there were no intergroup differences in terms of pain, edema, range of motion, and fitness for work. Bone scintigraphy was also similar. In 66 patients diagnosed with wrist or ankle CRPS, Gobelet et al.5 combined an intensive physiotherapy regimen (eight-week course) to a three-week course of nasal calcitonin or nasal placebo spray. At eight weeks, these authors noted decreased static and dynamic pain scores (using a four-point pain scale) in the treatment compared with the placebo group (0.45 ± 0.68 vs 0.69 ± 0.93; P < 0.007 and 0.77 ± 0.76 vs 1.22 ± 0.91; P < 0.04, respectively). Furthermore, range of motion was also greater in patients receiving calcitonin (P < 0.04). However, there were no differences between the two groups in terms of edema and ability to work. In 2006, Sahin et al.6 enrolled 35 patients with upper extremity CRPS and provided them with an exercise and physical therapy program. In addition, the subjects were randomized to a two-month regimen of nasal calcitonin or oral paracetamol. At two months, Sahin et al. 6 observed no intergroup differences pertaining to pain, range of motion, allodynia, hyperalgesia, and trophic changes.

Biphosphonates

Bone demineralization often accompanies CRPS. This observation has prompted many authors to advocate treatment with biphosphonates, which are potent inhibitors of bone resorption.7 To date, four RCTs have investigated the role of bisphosphonates in the treatment of CRPS.

In a study by Adami et al.,7 20 patients were randomized to receive a three-day course of intravenous alendronate or placebo. After two weeks of treatment, patients receiving alendronate presented significant improvements in pain, swelling, and range of motion compared with the control group (all P < 0.01). In light of these encouraging results, Manicourt et al.8 randomized 39 patients with lower limb CRPS to an eight-week regimen of oral alendronate or placebo. Participants were excluded if they had received calcitonin one week prior to study entry; furthermore, they were encouraged to continue physical therapy and rehabilitation on a regular basis. At four, eight, and 12 weeks, participants receiving alendronate displayed better pain control and range of motion compared with controls (both P < 0.05). Moreover, edema was also improved with alendronate at four and eight weeks (both P < 0.05). In 2000, Varenna et al. 9 randomized 31 patients with CRPS to a ten-day course of intravenous clodronate or placebo. To assess efficacy, the authors used a visual analogue scale of pain (VAS, range 0-100), clinical global assessment (CGA, range 0-3), and an efficacy verbal score (EVS, range 0-3). Forty days after treatment, subjects receiving clodronate displayed significantly improved VAS (22.3 ± 20.2 vs 56.4 ± 31.4; P < 0.001), CGA (0.9 ± 0.6 vs 1.9 ± 0.7; P < 0.001), and EVS scores (1.6 ± 0.7 vs 0.2 ± 0.4; P < 0.001) compared with controls. In 2004, Robinson et al. 10 randomized 27 patients to a single intravenous dose of palmidronate 60 mg or placebo. At the three-month evaluation, subjects who had received biphosphonates reported lower pain scores (P = 0.043), a greater overall improvement (P = 0.026), and higher functional assessment scores pertaining to physical function (P = 0.047).

Free radical scavengers

An excessive inflammatory reaction can lead to the overproduction of free radicals, resulting in the destruction of healthy tissue and possibly leading to CRPS. Thus, free radical scavengers have been proposed to curtail the disease process.11 To date, three free radical scavengers (dimethyl sulfoxide, N-acetylcysteine [NAC], and mannitol) have been investigated for the treatment of CRPS.

In 1996, Zuurmond et al. 11 randomized 31 patients to a daily application of a fatty cream containing 50% dimethyl sulfoxide (DMSO) or a placebo. In addition, all subjects received physiotherapy. After two months of treatment, the authors observed that patients in the study group displayed a greater improvement in the Reflex Sympathetic Dystrophy score (based on the presence of pain, edema, decrease range of motion, altered colour and temperature, as well as use-dependent worsening of symptoms) (3 vs 4; P < 0.01). However, the reductions in subjective pain were similar for the two groups. In 26 patients with a new diagnosis for CRPS (<three months), Geertzen et al. 12 compared a three-week course of 50% DMSO to intravenous regional blockade with ismelin (twice a week for three weeks). Using a scoring system based on pain, disability, edema, colour, and range of motion, these authors observed a greater improvement at seven and nine weeks in subjects randomized to DMSO.

Subsequently, Perez et al. 13 compared 50% DMSO with NAC. One hundred twelve patients, who were recently diagnosed with CRPS (<one year) and did not undergo prior sympathectomy, were randomized to a 17-week course of 50% DMSO or NAC. All subjects also received paracetamol, naproxen, tramadol, and occupational and physical therapy. The primary outcome measurement was the Impairment Level SumScore (ISS), which incorporates pain, temperature, volume, and active range of motion of the affected extremity. Perez et al.13 also assessed the effect of treatment on the disability level and the quality of life. After 17 weeks, no differences were found between the two groups. However, subgroup analysis revealed that patients with cold CRPS, defined as a –0.4°C difference between the affected and healthy extremity, showed a greater improvement in ISS with NAC compared with DMSO (P = 0.04). In contrast, subjects with warm CRPS, defined as a +0.4°C difference between the affected and healthy extremity, displayed more improvement in the quality of life with DMSO (P = 0.001). Follow up at 52 weeks again revealed no major differences between the two groups.

van Dieten et al.14 performed a cost analysis using a pharmacoeconomic evaluation conducted in parallel to Perez et al.’s trial. Total costs were defined as the sum of direct costs within the healthcare system (visits to healthcare providers, prescribed medication, occupational devices, and home care), direct costs outside the healthcare system (travel expenses, costs of alternative treatment, over-the-counter medication, and family care), and indirect costs (absenteeism, loss of productivity). Over the course of the 52 weeks, van Dieten et al.14 observed that DMSO resulted in lower total direct costs than NAC (2852 € vs 3934 €; P < 0.05).

Mannitol has also been investigated in the treatment of CRPS. Perez et al.15 randomized 41 patients to receive 10% mannitol intravenously or placebo, each to be administered on five consecutive days. These authors found no intergroup differences in pain (assessed daily during nine weeks). Furthermore, the levels of impairment, disability, and handicap were also similar between the two groups at two, six, and nine weeks.

Steroids

Biopsy studies showing tissue inflammation in CRPS have led many authors to use steroids.16 To date, three RCTs have investigated the role of steroids in the treatment of CRPS.

Christensen et al.16 randomized 23 patients with upper extremity CRPS to prednisone or placebo. The medication was continued until a clinical response was obtained, but for no more than 12 weeks. Using a scale based on pain, edema, volar sweating, and finger-knitting ability, the authors observed 75% improvement rates of 100% and 20% in the treatment and control groups, respectively (P < 0.01). Braus et al.17 enrolled 34 patients suffering from upper limb CRPS secondary to cerebral infarct. The subjects were randomized to a four-week course of methylprednisolone or placebo. After treatment, the steroid group presented a decreased shoulder-hand syndrome (SHS) score (based on pain/hyperalgesia, distal edema, and passive humeral abduction/external rotation). When patients in the placebo group were crossed over to the treatment arm, these benefits persisted. Furthermore benefits were still present at six months (SHS score < 3/14). Using a similar population, Kalita et al.18 randomized 60 patients to a five-week course of prednisolone or piroxicam. After treatment, the prednisolone group displayed lower SHS scores (4.27 ± 2.83 vs 9.37 ± 2.89; P < 0.001). In both groups, no changes were detected in activities of daily living.

Gabapentin

Because neuropathic pain can be a prominent feature in CRPS, gabapentin, an anticonvulsant with proven analgesic effect in various neuropathic pain syndromes, has been investigated.19

van de Vusse et al.19 undertook a double-blind randomized crossover study in 46 patients with long standing CRPS that was refractory to sympathetic blocks, mannitol infusions, and transcutaneous modulation. At first, the subjects were randomized to receive a three-week course of gabapentin or placebo. This was followed by a two-week washout period. Subsequently, the patients were crossed over. Overall, there were no differences in pain scores between treatment and control groups. However, using global perceived pain relief, more patients receiving gabapentin reported an improvement in pain control (43% vs 17% of patients; P = 0.002). Subjects receiving gabapentin also reported more side effects (dizziness, somnolence, lethargy) (all P ≤ 0.003).

Tadalafil

During the chronic phase of CRPS, impaired microcirculation can lead to tissue hypoxia and metabolic tissue acidosis. Tadalafil is a vasodilator that inhibits phosphodiesterase 5, used to treat erectile dysfunction and pulmonary arterial hypertension.20

Groeneweg et al. 20 randomized 24 patients suffering from cold CRPS of a lower limb to a 12-week course of placebo or tadalafil. In addition, all subjects continued physiotherapy. After treatment, patients in the tadalafil group experienced a greater reduction in VAS scores (15% vs 0%; P = 0.004). However temperature changes, muscle strength, and activity level were similar between the two groups.

Sarpogrelate hydrochloride

Sarpogrelate hydrochloride, a selective 5-HT2 antagonist, has been shown to improve peripheral blood circulation through inhibition of serotonin-induced platelet aggregation and vasoconstriction.21

Ogawa et al.21 randomized 30 patients with CRPS to conventional treatment (sympathetic blocks, analgesics, antiepileptics, antidepressants, sedatives, physical therapy) or a three-month course of sarpogrelate combined with conventional therapy. At the end of treatment, no differences in pain were observed between the two groups. However, with sarpogrelate, a greater proportion of patients reported improvement in burning pain sensation (70% vs 0%; P < 0.05).

Interpretation

In all placebo-controlled RCTs, biphosphonates have been shown to decrease pain and swelling as well as to increase range of motion for patients with CRPS. In most trials pertaining to calcitonin, benefits associated with its administration were not detected. The effect of free radical scavengers may be drug dependent; while mannitol is no better than placebo, DMSO seems to provide a mild improvement in range of motion and vasomotor instability in patients with CRPS. Owing to its costs, NAC is best reserved for a subgroup of patients with cold CRPS. A short course of oral steroids (prednisolone or methylprednisolone) may help with pain control, edema, and mobility in CRPS patients with or without cerebral infarcts. In light of the marginal benefits or limited supportive evidence, tadalafil, sarpogrelate, and gabapentin should be used with caution.

Intravenous regional blockade and nerve blocks

Intravenous regional blockade (IVRB)

Since CRPS has been traditionally associated with a dysfunction of the sympathetic nervous system, many authors have advocated treatment with sympathetic blockade, achieved using an IVRB consisting of local anesthetics, guanethidine (an inhibitor of the presynaptic release of norepinephrine), reserpine (an agent inhibiting norepinephrine synthesis and depleting norepinephrine stores), and/or droperidol (an alpha adrenergic antagonist). Ketanserin (a serotonin type 2 receptor antagonist) and atropine have also been used, although the beneficial effects with these agents were thought to occur through neuromodulation.

Currently, there are seven RCTs comparing IVRB therapy with placebo; another four RCTs compare various therapeutic agents for IVRB.

Randomized controlled trials comparing treatment with placebo

In 1990, Blanchard et al.22 randomized 21 patients with CRPS to IVRB with guanethidine, reserpine, or normal saline (NS). Both guanethidine and reserpine were diluted in NS. Pain scores measured at weekly intervals during 12 weeks were not significantly different between the three groups. Subsequently, Jadad et al.23 enrolled nine patients with CRPS who were known to be responsive to sympathectomy with guanethidine. In each subject, one low dose of guanethidine (10 and 20 mg for the upper and lower limb, respectively) diluted in NS, one high dose of guanethidine (30 mg for both upper and lower limbs) diluted in NS, and one dose of plain NS were tested on three occasions separated by intervals of one week. The order of the solutions was random. Jadad et al.23 observed no intergroup differences in terms of pain intensity and relief, mood, and duration of analgesia. In 2002, Livingstone et al.24 randomized 56 patients with upper extremity CRPS to IVRB with NS or guanethidine (diluted in 0.5% prilocaine). Based on the clinical response, further blocks, to a maximum of four, were administered at weekly intervals. Assessments (pain, vasomotor instability, digital tenderness, and stiffness) were carried out at 24 hr, 48 hr, and one week after each block. Compared with the control group, Livingstone et al. 24 found no benefits associated with guanethidine. In fact, long term analysis at 15 weeks revealed that patients receiving guanethidine were more likely to have persistent alterations in hand color (P = 0.015); the colour and temperature of their hands also exhibited more sensitivity to ambient thermal changes (P = 0.003). Moreover, at 30 weeks, more subjects complained of altered hand temperature (69% vs 14%; P < 0.001) and digital swelling (P = 0.04).

Although guanethidine is the most commonly used drug for IVRB, some authors have investigated alternative agents, such as droperidol, ketanserin, atropine, and methylprednisolone. Kettler et al. 25 enrolled six patients with CRPS who had previously obtained a significant but transient relief with a local anesthetic sympathetic block. The subjects were randomized to IVRB with droperidol/heparin/NS or placebo (heparin/NS). Pain was monitored daily. After two weeks, the procedure was repeated with the alternate solution. Droperidol was not associated with any analgesic benefits; however, in three patients, akithesia, dysphoria, or nausea occurred. Hanna et al.26 randomized nine patients to a series of two IVRB with ketanserin followed by two IVRBs with placebo or vice versa. Treatments were provided at weekly intervals. No differences in pain scores were seen. However, ketanserin was associated with more drowsiness, faintness, and shakiness (all P < 0.05). In 1993, Glynn et al. 27 recruited 14 patients with CRPS that was confirmed by pain relief after a guanethidine IVRB. The subjects were randomized to IVRB with atropine diluted in NS or NS alone. Patients could receive up to two treatments at a weekly interval before crossing over to the alternate solution. There were no differences in pain, pain relief, and mood. Taskaynatan et al.28 recruited 22 patients with upper limb CRPS who did not undergo sympathetic blockade in the prior month. Treatment was randomized to IVRB with methylprednisolone and 2% lidocaine (diluted in NS) or NS. A series of three treatment sessions was carried out at weekly intervals. Pain, range of motion, edema, and patient satisfaction, measured on each occasion one hour after tourniquet release, revealed no differences between the two groups. These variables were still similar at 1.5 months.

Randomized controlled trials comparing therapeutic modalities

Rocco et al.29 recruited 12 patients with CRPS that presented temporary relief with either stellate or lumbar sympathetic block. IVRB was achieved using guanethidine diluted in 0.5% lidocaine, reserpine diluted in 0.5% lidocaine, or plain 0.5% lidocaine. Every subject underwent treatment with all three solutions in a randomized fashion at weekly intervals. Pain scores were evaluated over 90 min after each block, and patients maintained an hourly pain log during the week following treatment. There were no differences between the three groups.

In 1995, Ramamurthy et al. 30 attempted to determine the optimal number of blocks for an IVRB with guanethidine. Fifty-seven patients with a recent history of CRPS (<three months), who had not undergone prior IVRB, were randomized to receive one, two, or four IVRBs. The blocks were carried out at four-day intervals using guanethidine diluted in 0.5% lidocaine. Pain, global evaluation (perceived improvement), and range of motion measured after each session and at one, three, and six months, were similar after one, two, or four blocks.

In 1983, Bonelli et al.31 compared IVRB and stellate ganglion blockade. Nine patients were randomized to a series of four IVRBs with guanethidine performed every four days or to a sequence of eight stellate ganglion blocks (0.5% bupivacaine) carried out every other day. Although IVRB produced a greater increase in skin temperature and plethysmographic tracing 24 and 48 hr after the blocks, no intergroup differences in pain were observed at 15 min, one hour, 24 hr, 48 hr, 16 days, one month, and three months. Furthermore, at one and three months, there were no differences in vasomotor disturbances, trophic changes, edema, and range of motion.

In addition to guanethidine and reserpine, bretylium has also been used to treat CRPS. In 1992, Hord et al.32 enrolled seven patients who had a history of temporary relief with stellate ganglion or sympathetic blocks. Each subject received two IVRB with bretylium (diluted in 0.5% lidocaine) and two IVRB with lidocaine. The sequence of treatments was randomized. A > 30% decrease in pain was considered clinically significant; therefore, when the patient’s pain relief was less than 30%, the next IVRB was performed. Hord et al. 32 observed that IVRB with bretylium and lidocaine provided a longer analgesic duration than pure lidocaine (20 ± 17.5 days vs 2.7 ± 3.7 days; P < 0.001).

Stellate ganglion block (SGB) and lumbar sympathetic block (LSB)

The stellate and lumbar sympathetic ganglia are responsible for the sympathetic innervation of the upper and lower limb, respectively. Many authors have sought to interrupt these sympathetic pathways through local anesthetic blockade, chemical neurolysis, and/or radiofrequency neurotomy.

Price et al.33 recruited four patients with upper extremity CRPS and three patients with lower extremity CRPS for whom they performed SGB and LSB, respectively. Each subject received one block with local anesthetics (1% lidocaine and 1% lidocaine/0.25% bupivacaine for SGB and LSB, respectively) and one with NS. The two blocks were separated by an interval of seven to ten days; the order was random. Thirty minutes after the injection, Price et al.33 found no differences between the peak analgesic effects of the two groups. The duration of pain relief was greater with local anesthetics than NS (five days and 12 hr vs six hours; P < 0.02).

In 2008, Manjunath et al.34 set out to compare phenol neurolysis and thermal radiofrequency of the lumbar sympathetic chain. Nineteen patients with lower limb extremity CRPS lasting more than six months, refractory to oral medications (gabapentin, amitryptiline, carbamazepine) but responsive to sympathetic blocks, were randomized to receive thermal radiofrequency or neurolysis with 7% phenol. Despite a reduction in pain scores compared with baseline, no intergroup differences were observed; follow up was carried out until four months. In a previous RCT (n = 17), Haynsworth et al.35 also compared thermal radiofrequency (TRF) with phenol neurolysis for patients with lower extremity CRPS. The authors did not record the analgesic efficacy of either technique; however, they observed that sympathectomy (assessed by temperature and sweat test) was still present at eight weeks in 89% of patients in the phenol group compared with 12% in the TRF group (P < 0.05). The incidence of postsympathectomy neuralgia (11-33%) was not statistically different.

By preventing the release of acetylcholine from preganglionic sympathetic nerves, botulinum toxin type A (BTA) has been shown to induce a prolonged sympathetic block in animals.36 Carroll et al. 36 recruited seven patients with lower limb CRPS lasting more than six months, refractory to two nonopioid medications but transiently responsive to a previous lumbar sympathectomy. Each subject received two LSBs consisting of bupivacaine or a combination of bupivacaine and BTA. The order was random and patients were eligible for the crossover injection one month after the disappearance of pain relief. Carroll et al.36 found a longer duration of analgesia in the bupivacaine-BTA group (71 days vs < ten days; P < 0.002).

When performing LSB prior to LA injection, the needle tip’s position is commonly verified with fluoroscopy and contrast agents. Tran et al. 37 set out to investigate if the latter affected the outcome of the block. Fifteen patients undergoing LSB for CRPS were randomized to receive iohexol or saline prior to the injection of LA. Over the course of the following week, these authors found no major differences between the two groups in terms of pain, allodynia, percent of relief from pain, and interference with daily activities.

Epidural clonidine

By reducing the sympathetic nervous activity, α2-adrenergic agonists administered in the epidural space may contribute to decreased pain in patients with CRPS.38

In 1993, Rauck et al.38 enrolled 26 patients with CRPS that was no longer responsive to sympathetic blocks. Cervical (C7-T1) and lumbar (L2-3) epidural catheters were inserted for upper and lower extremity CRPS, respectively. On consecutive days, patients received an epidural injection of clonidine 300 μg, clonidine 700 μg, or NS in random order. Throughout the study period (six hours), the authors found that pain (measured by VAS and the McGill Pain Questionnaire) was significantly improved in both treatment groups compared with placebo. There were no differences in analgesia between the two doses of clonidine. However, sedation scores were higher in patients receiving 700 μg (P < 0.001).

Intrathecal baclofen

The intrathecal administration of baclofen, a γ-aminobutyric acid – receptor (type B) agonist that inhibits sensory input to the spinal cord, has proven beneficial to some patients with dystonia.39 Since CRPS can lead to dystonia and is often unresponsive to standard treatment, baclofen has been tried in CRPS.

In seven patients with dystonia, refractory to benzodiazepines, levodopa, antiepileptic drugs, botulinum toxin, mannitol, surgical/chemical sympathectomy, and oral baclofen, van Hilten et al. 39 inserted an intrathecal catheter and placed its tip at the T11-T12 level. Each subject received a daily bolus of baclofen (25, 50, or 75 μg) or NS in a randomized fashion. The authors observed that the 50 and 75 μg doses provided a significant decrease in dystonia compared with NS and 25 μg (P < 0.05).

Interpretation

In the setting of IVRB, the available evidence does not support the use of guanethidine, reserpine, droperidol, ketanserin, atropine, or lidocaine-methylprednisolone. Compared with placebo, local anesthetic agents prolong the duration of action but do not alter the peak effect of SGB and LSB. Two small studies (n = 7) suggest that combining local anesthetic agents with bretylium or botulinum toxin can increase the analgesic duration of IVRB and LSB, respectively. For LSB, although no analgesic differences were found between TRF and phenol neurolysis, the latter may result in longer lasting sympatholysis. Epidural clonidine (300 μg) and intrathecal clonidine (50-75 μg) constitute interesting alternatives for refractory lower limb CRPS and CRPS-related dystonia, respectively. However these neuraxial drugs require further investigation.

Spinal cord stimulation (SCS)

Spinal cord stimulation requires that an electrode be surgically placed in the epidural space at the level of the nerve roots innervating the painful area. An electrical current from the electrode induces paresthesiae, a sensation that suppresses pain. The current is supplied by a pulse generator that is located subcutaneously in the anterior abdominal wall. Patients can subsequently customize the intensity of the current by means of a device that uses radio-frequency transmission.40 One RCT by Kemler et al.40 studied the use of SCS in CRPS, while two subsequent studies from the same authors looked at the two- and five-year follow ups.41,42

Kemler et al. 40 recruited 54 patients suffering from CRPS for more than six months who had failed conventional treatment. Participants were randomized to SCS combined with standardized PT or PT alone. Allocation was done on a 2:1 basis in favour of the SCS group. For the latter, subjects first underwent trial lead placement. Those obtaining more than 50% improvement on VAS or more than six points on a seven-point global perceived effect (GPE) scale received permanent lead placement. The main outcome variables were the differences between pre-randomization measurements and those at six months for the VAS, GPE, functional measurements, and quality of life indicators. Thirty-six patients were recruited to the SCS/PT group; 24 proceeded to permanent lead placement. Eighteen patients were allocated to the PT only group. Using an intent-to-treat analysis, changes in VAS favoured the SCS/PT group at six months (−2.4 ± 2.5 vs +0.2 ± 1.4; P < 0.001). Furthermore, the percentage of patients achieving a GPE score of 6/7 or higher was 36% for the SCS/PT group and 6% for the PT group (P = 0.01). No differences were found for the changes in functional status and quality of life.

At two years, the SCS/PT group continued to have statistically better results (VAS and GPE).41 However, at five years, there were no statistical differences in any of the measured variables.42

Over the five-year study period, 42% of patients with SCS experienced at least one complication, with 72% of adverse events occurring during the first two years. The most common complication was pulse generator failure, followed by lead displacement and need to revise the pulse generator pocket.40-42

Interpretation

In patients with CRPS, SCS can provide significant pain reduction for up to two years. However, this effect may be lost between the second and fifth year of treatment. Furthermore SCS appears to have no impact on functional status or quality of life. During the first two years, SCS is also associated with a 72% rate of complications (most commonly generator failure, lead displacement, and need to revise the pulse generator pocket). Thus, in light of the adverse events and costs, further RCTs are needed to support the use of spinal cord stimulation.

Adjuvant therapy

Physiotherapy (PT) and occupational therapy (OT)

In 1999, Oerlemans et al.43 set out to investigate the role of PT and OT in the treatment of CRPS. They recruited 135 patients suffering from upper limb CRPS, which was present for less than one year and had not been treated with prior sympathectomy. After receiving free radical scavengers (DMSO or NAC), vasodilators (verapamil, ketanserine, or pentoxifylline), and trigger point injections, subjects were randomized to PT, OT, or control therapy (CT). The treatment sessions were continued until no further progress could be achieved or until their cessation did not result in a recurrence of symptoms. Their intensity and frequency were adjusted to individual patient needs. Goals for PT included improving pain control and increasing coping mechanisms. For OT, the authors tried to reduce the inflammatory symptoms, support the limb in the most functional position, normalize sensibility, and improve function for daily activities. After one year, Oerlemans et al. 43 observed that there were no significant differences in the decrease of pain (VAS scores) for all three groups. However, using a per-protocol analysis, scrutiny of the dimensions of pain (as assessed by the McGill Pain Questionnaire, Dutch Language Version) revealed that patients in the PT group selected fewer total and sensory words to describe their pain compared with their OT counterparts (all P values < 0.05). Compared with OT and CT, PT also provided improved active range of motion of the thumb at one year (P < 0.05).

Subsequently, Oerlemans et al. 44 calculated the impairment rating in all three groups one year after their inclusion in the study. The impairment rating was performed according to the American Medical Association’s Guides to the Evaluation of Permanent Impairment and included active range of motion, two-point discrimination, and grip strength. No significant differences were found between the three groups.

In the third publication based on the collected data, Oerlemans et al. 45 then set out to compare the levels of impairment, disability, and handicap. In terms of impairment, the authors observed no differences in ISS at one year. In terms of disability, the Raboud Skills Questionnaire and modified Greentest could not detect any differences between the three groups. The level of handicap, assessed by sickness impact profile (SIP) scores, was also similar.

In the final study, the same group of authors proceeded to calculate the cost-effectiveness of adjuvant PT and OT.46 Total calculated costs included medical, non medical, and loss of productivity costs. The incremental cost-effectiveness ratios (ICER) were calculated by first obtaining the mean difference in effectiveness between end-of-study and baseline measures. The total medical costs were then divided by the latter. When the authors examined the ICER for SIP, both PT and OT proved to be more costly and less effective than CT. For the Greentest, compared with CT, OT yielded an ICER of 992 Netherland guilders (NLG) per point of effectiveness; again, PT was more costly and less effective. Physiotherapy and OT both provided an improvement in ISS compared with controls, and PT was more cost efficient than OT (184 NLG/point vs 1152 NLG/point, respectively).

The optimal frequency of PT sessions was examined by Lee et al. 47 The authors recruited 28 children, aged eight to 17 yr and suffering from lower extremity CRPS, who had not received sympathetic blocks or more than two sessions of physiotherapy. Cognitive-behavioural therapy (relaxation training, deep breathing exercises, biofeedback, and guided imagery) was provided to all subjects. They were subsequently randomized to receive PT at a frequency of one or three times a week for six weeks. The PT program was individualized to each patient and included transcutaneous electrical nerve stimulation, progressive weight bearing, tactile desensitization, massage, as well as contrast baths. At the short term (six weeks to three months after treatment) and long term follow ups (six to 12 months after treatment), no intergroup differences were noted in terms of outcomes related to pain and functionality. Furthermore, a phone interview (conducted at a mean of 133 weeks) also revealed no differences pertaining to pain, function (ambulation), CRPS recurrence, activity level (participation in sports), and school attendance.

Motor imagery program (MIP)

Moseley48 recruited 13 patients who developed CRPS after a wrist fracture and randomized them to a MIP or to continue their ongoing (conventional) treatment. The MIP incorporated recognition of hand laterality (subjects were presented with pictures of right and left hands and asked to identify the correct side), imagined hand movement (subjects were presented pictures of a hand in different positions and asked to imagine moving their own hand to adopt the posture shown), as well as mirror movements (subjects placed both hands into a box with a mirror separating the two compartments and, while moving both hands, were asked to watch the reflection of the unaffected hand in the mirror). Each stage lasted two weeks, and subjects were required to perform the tasks hourly from 8 a.m. to 8 p.m. No restrictions were placed on the conventional treatment. Six and twelve weeks after the completion of treatment, patients in the MIP group presented less pain and decreased swelling (both P < 0.05). The beneficial effects of treatment were replicated when the controls crossed over to the MIP group.

In a follow-up study, Moseley49 randomized 20 patients to receive MIP in three different sequences: recognition of laterality/imagined movements/mirror movements (RecImMir), imagined movements/recognition of laterality/imagined movements (ImRecIm), and recognition of laterality/mirror movements/recognition of laterality (RecMirRec). At 12 weeks, the author found that the RecImMir group experienced a greater decrease in pain and an increase in functionality compared with the other two groups (both P < 0.05).

Interpretation

Due to contradictory results, the benefits pertaining to PT and OT remain unclear. In adults, the optimal frequency of PT has not been determined. In children, weekly and triweekly PT sessions yield similar results when combined with cognitive-behavioural therapy. The optimal type of PT has not been elucidated. Motor imagery seems to offer promising results but further trials are required to validate its use.

Limitations

For practical reasons, a decision was taken to limit this review to RCTs published in the English language. Although such a restriction may constitute a methodological limitation, we believe that its impact on the paper’s conclusions is small. Expansion of our search criteria (using the same databases and time periods) to languages other than English only yielded an additional eight RCTs.50-57 In this review, no attempt was made to produce a meta-analysis. In our view, given the wide array of therapeutic modalities used for CRPS, patient enrolment would have been insufficient for many treatments to support a systematic pooling of data. We also made no distinction between CRPS type 1 and type 2; except for the documented presence of nerve injury, both entities appear to be clinically similar.2 Finally, no RCT was excluded based on factors such as sample size justification, statistical power, blinding, definition of intervention allocation, or clinical outcomes. This may represent a limitation to our review, as it may serve to overemphasize evidence derived from “weaker” RCTs. Most importantly, we cannot exclude the possibility that trials that lacked sample size justification, provided limited enrolment, and found no difference between study groups were inadequately powered to answer the question they sought to investigate.

Conclusions

In terms of pharmacological treatment, oral and intravenous biphosphonates, but not calcitonin, have been proven to reliably decrease pain and swelling as well as increase range of motion in patients with CRPS. For free radical scavengers, topical DMSO constitutes the best option. However, its beneficial effects appear to be mild. A short course of oral steroids may be indicated for CRPS patients with or without cerebral infarcts. Limited benefits or supportive evidence suggest that tadalafil, sarpogrelate, and gabapentin should be used with caution.

In the setting of IVRB, the available evidence does not support the use of guanethidine, reserpine, droperidol, ketanserin, atropine, or lidocaine-methylprednisolone. Compared with placebo, local anesthetic agents prolong the duration of action but do not alter the peak effect of stellate and lumbar sympathetic blocks. Epidural and intrathecal clonidine constitute interesting alternatives for refractory lower limb CRPS and CRPS-related dystonia, respectively, but these neuraxial drugs require further investigation. Two small studies suggest that combining local anesthetic agents with bretylium or botulinum toxin can increase the analgesic duration of IVRB and LSB, respectively.

Considering the high incidence of side effects, limited analgesic duration, and costs, further studies are required to support the use of spinal cord stimulation. The benefits derived from adjunctive PT and OT remain unclear. Further trials are required to validate the use of motor imagery programs.

In summary, only biphosphonates appear to offer clear benefits for patients with CRPS. Improvement has been reported with dimethyl sulfoxide, steroids (in CRPS with or without cerebral infarcts), epidural clonidine, intrathecal baclofen, spinal cord stimulation, and motor imagery programs; further trials are required to confirm these findings. The available evidence does not support the use of calcitonin, vasodilators, or sympatholytic and neuromodulative intravenous regional blockade. Clear benefits have not been reported with stellate/lumbar sympathetic blocks, mannitol, gabapentin, and physical/occupational therapy, and further studies are required (Table 5).

Table 5 Randomized controlled trials comparing treatment with control

Caution should be exercised when interpreting the limited data available in the literature on CRPS. Many issues regarding these therapeutic modalities remain unresolved; thus, they require elucidation through well-designed and meticulously conducted RCTs. Future trials should use uniform diagnostic criteria for CRPS; sample size justification and blinded assessment should be systematically implemented. Furthermore, the duration of CRPS prior to enrolment and the length of follow up need to be rigorously controlled. Study endpoints should include pain relief but also reversal of trophic changes and improvement of functionality (range of motion). Lastly, most studies have thus far focused on single or dual therapeutic modalities; the role of multimodal therapy warrants investigation.