Complex Regional Pain Syndromes

Updated: Jun 20, 2018
  • Author: Gaurav Gupta, MD, FAANS, FACS; Chief Editor: Stephen A Berman, MD, PhD, MBA  more...
  • Print


In 1994, a consensus group of pain medicine experts gathered by the International Association for the Study of Pain (IASP) agreed on diagnostic criteria for reflex sympathetic dystrophy (RSD) and causalgia, and renamed them complex regional pain syndrome (CRPS) types I and II, respectively. These designations were determined by the type of inciting event, rather than by any differences in clinical presentation or pathophysiology. Many experts felt that the IASP diagnostic criteria were ambiguous; however, these criteria were developed as just a starting point, and the IASP fully intended to validate them through clinical research studies. [1]

CRPS type I requirements feature causation by an initiating noxious event, such as a crush or soft tissue injury; or by immobilization, such as a tight cast or frozen shoulder. CRPS type II is characterized by the presence of a defined nerve injury. Both types demonstrate continuing pain, allodynia, or hyperalgesia that is usually disproportionate to the inciting event. At some point during the syndrome's development, both show evidence of edema, changes in skin blood flow revealed by color changes and skin temperature changes greater than 1.1°C from the homologous body part, or abnormal sudomotor activity in the painful region. Both types require the exclusion of any other condition that might account for the degree of pain and dysfunction seen. [2]

The 1994 IASP criteria have proven to be extremely sensitive (ie, they rarely miss a true case of CRPS). However, since their inception, the 1994 taxonomy has been criticized by experts on clinical criteria validation and specialists in pain medicine on the grounds that the criteria are insufficiently specific (ie, use of the criteria results in overdiagnosis of CRPS). A small single center validation study demonstrated empirically that the 1994 CRPS criteria did indeed cause overdiagnosis of the syndrome. [2]

In response to such concerns, investigators used factor analysis to categorize 123 patients with CRPS into 4 statistically distinct subgroups. [3] This resulted in modified diagnostic criteria felt to be valuable for further validation studies. [3, 4]

In 2003, a closed workshop was held in Budapest, Hungary to study and resolve this matter. Experts in CRPS published the results of this workshop in a 2007 review article [5] showing that the modified criteria, mentioned above, produced better discrimination between CRPS and non-CRPS neuropathic pain, yielding better diagnostic accuracy than the original unmodified criteria. [5]

The study results indicated that when 2 of 4 sign categories were present and 3 of 4 symptom categories were present, the resultant sensitivity was 0.85 and the specificity was 0.69 for a clinical diagnosis of CRPS. This appeared to be a good compromise between identifying as many patients as possible in the clinical context and substantially reducing the high level of false-positive diagnoses associated with the 1994 IASP criteria. However, a higher specificity is required to meet research criteria, so the committee recommended that 2 of the 4 sign categories and all 4 symptom categories must be positive for the diagnosis to be made in a research setting, resulting in a sensitivity of 0.70 and specificity of 0.94.

Due to the combination of increased specificity and reduced sensitivity, about 15% of patients previously diagnosed with CRPS were considered "without a diagnosis." Therefore, a third diagnostic subtype, complex regional pain syndrome not otherwise specified (CRPS-NOS), was recommended to categorize those patients. [6] These new IASP diagnostic criteria have been submitted to the medical committee for Classification of Chronic Pain of the IASP for future revision of formal taxonomy and diagnostic criteria. 

The criteria are given here in hopes that higher specificity for the identification of CRPS will enhance research into the pathoetiology of this disorder without creating a reduced, or even harmful, rate of clinical diagnosis that could deny affected patients access to treatment. In addition, these criteria may result in more cost-effective approaches for the management of this disorder. [5, 7, 6]  These criteria, as listed below, are given in the most current version of the IASP's Complex regional pain syndrome: practical diagnostic and treatment guidelines (4th edition) [8]  where they are described as "state of the art" diagnostic criteria and "practical" guidelines. 

IASP-proposed revised CRPS clinical diagnostic criteria

A clinical diagnosis of CRPS can be made when the following criteria are met:

  • Continuing pain that is disproportionate to any inciting event

  • At least 1 symptom reported in at least 3 of the following categories:

    • Sensory: Hyperesthesia or allodynia

    • Vasomotor: Temperature asymmetry, skin color changes, skin color asymmetry

    • Sudomotor/edema: Edema, sweating changes, or sweating asymmetry

    • Motor/trophic: Decreased range of motion, motor dysfunction (eg, weakness, tremor, dystonia), or trophic changes (eg, hair, nail, skin)

  • At least 1 sign at time of evaluation in at least 2 of the following categories:

    • Sensory: Evidence of hyperalgesia (to pinprick), allodynia (to light touch, temperature sensation, deep somatic pressure, or joint movement)

    • Vasomotor: Evidence of temperature asymmetry (>1°C), skin color changes or asymmetry

    • Sudomotor/edema: Evidence of edema, sweating changes, or sweating asymmetry

    • Motor/trophic: Evidence of decreased range of motion, motor dysfunction (eg, weakness, tremor, dystonia), or trophic changes (eg, hair, nail, skin)

  • No other diagnosis better explaining the signs and symptoms

In addition, a slightly modified version of the above listing is used for CRPS research (as opposed to clinical) criteria. For these rules one must have the CRPS characteristics present in all four of the symptom categories and in at least two out of the four sign categories.



Hypothetical mechanisms

In most cases, experts believe that CRPS develops when persistent noxious stimuli from an injured body region leads to peripheral and central sensitization, whereby primary afferent nociceptive mechanisms demonstrate abnormally heightened sensation, including spontaneous pain and hyperalgesia. Allodynia and hyperalgesia occur when central nervous system (CNS) somatosensory processing misinterprets normal nonpainful mechanical stimuli, such as light touching of the skin, as painful. Therefore, skin in the injured area becomes more sensitive to all stimuli, even nonpainful stimuli. In addition, the sensitization can extend beyond the originally injured area, thus enlarging the region of aberrant pain perception.

A similar impairment of CNS processing leads to motor aberrancies, such as weakness or tremor in the affected area. The peripheral and central sensitization associated with impaired CNS processing is linked to proposed disturbances within the sympathetic nervous system (SNS) that lead to sympathetic hyperactivity adversely affecting the injured area. Studies suggest that an augmented inflammatory response coupled with impaired healing further contribute to the refractory nature of malevolent CRPS. [9, 1, 10, 11, 12, 13]

Peripheral and central sensitization

Mechanical, thermal, and chemical stimuli activate peripheral nociceptors that transmit pain messages through lightly myelinated A-delta fibers and unmyelinated C fibers projecting to Rexed layers I, II, and V in the spinal cord. This process leads to the release of excitatory amino acids, such as glutamine and asparagine, which then act upon N -methyl-D -aspartic acid (NMDA) receptors, causing the release of substance P (SP). SP then lowers the threshold for synaptic excitability in normally silent second-order interspinal synapses. [9, 1, 10, 11, 12]

Peripheral sensitization occurs when persistent or repetitive noxious stimulation of high-threshold, polymodal C fibers results in enhanced sensitivity, lower stimulus thresholds, and the prolonged, enhanced activation of dorsal horn cells, especially those with glutamate receptors. In addition to SP, algogenic substances that are typically involved in tissue damage and capable of inducing transduction centripetally include potassium, serotonin, bradykinin, histamine, prostaglandins, and leukotrienes. Neuropeptides, such as SP and calcitonin gene-related peptide (CGRP), are also transported to the endings of nociceptive afferents where they can instigate ortho- and retrograde actions including, but not limited to, neurogenic inflammation, which can incite a host of additional hostile algogenic mechanisms.

Chronic CNS sensitization is engendered through afferent processing by second-order nociceptor-specific neurons and wide-dynamic-range (WDR) neurons in the spinal cord. WDR neurons contribute more to sensitivity than nociceptor-specific neurons, because both nociceptive and non-nociceptive afferents converge to synapse on a single WDR neuron, and WDR neurons respond with equal intensity regardless of whether the neural signal is noxious (hyperalgesia) or not.

Hyperalgesia and allodynia initially develop at the injury site. However, after CNS sensitization occurs through WDR neural activity, the area of pain expands beyond the initial region of tissue pathology. The peripheral changes described eventually cause an injury environment, where primary afferents, including nociceptors, demonstrate an increased sensitivity to circulating or experimentally injected subcutaneous norepinephrine. [14, 1, 10, 11, 12, 15, 16, 17, 18, 19, 20]

In addition to functional CNS sensitization, recent investigations have explored the possibility that the brain of those with CRPS may differ structurally from brains of those without CRPS. Studies of small numbers of patients from various authors suggest that patients with CRPS may have, for example, diminished thickness of the prefrontal cortex [21]  or diminished gray matter volume in certain regions related to pain perception but greater gray matter volume in other regions. [22]  Another group found evidence that the choroid plexus is enlarged in CRPS. [23]   More recently, van Velzen et al. studied 19 of their own patients and conducted a review of the existing literature. They failed to find any specific differences in the structure of function of the brain in CRPS patients and they also concluded that prior results in the literature were inconsistent with regard to the location and amount of the supposed changes as well as in the direction of the changes. [24] At present, the idea that structural brain changes may underly CRPS remains intriguing but inconclusive.

Sympathetically maintained pain (SMP)

For decades, CRPS was thought to be caused by SNS hyperactivity, and the pain experienced by those who suffer from CRPS was believed to be SMP. SNS involvement in CRPS is supported clinically by the presence of abnormal patterns in skin temperature, skin color, and sweating in the affected extremities. Surgical and chemical sympathectomy can relieve pain in some cases. However, under normal physiological circumstances, there is no interaction between the sympathetic and peripheral afferent nociceptive neurons. [14, 1, 25, 26, 27] Furthermore, multiple discrepancies undermine the possibility of SNS involvement. These discrepancies include the following: (1) plasma catecholamine concentrations are lower in CRPS-affected limbs [28, 29] , (2) most CRPS patients do not obtain significant or lasting pain relief from sympathetic blocks [30, 31] , and (3) skin temperature does not correlate with the activity of sympathetic vasoconstrictor neurons. [32]

To explain these incongruities, the pathophysiology of SMP was hypothesized to involve an abnormal coupling between sympathetic efferent and nociceptive afferent neurons. [33] Two possible conditions may lead to pathological coupling: interactions between sympathetic efferents and intact or regenerating peripheral nociceptive C-fiber neurons, or between sympathetic vasoconstrictor neurons and afferent somata within the dorsal root ganglion (DRG). [34]

This coupling is mediated by norepinephrine, which is released from newly expressed sympathetic terminals and adrenoreceptors onto afferent nociceptive neurons. Indeed, increased mRNA for alpha-2-adrenoreceptors has been demonstrated in DRG neurons following a nerve injury. [35] Therefore, an increased number of targeted and functionally upregulated adrenoreceptors on lesioned nociceptive afferents, which has been demonstrated, would explain how reduced SNS activity in CRPS is capable of maintaining pain. [26, 1]

Evidence suggests that early autonomic symptoms and signs of CRPS are indicative of CNS dysfunction. [36] Wasner et al suggest that warmth of the affected extremity in the early stages of CRPS I is caused by the functional inhibition of central cutaneous vasoconstrictor activity, leading to cutaneous vasodilation. [27] However, over time, this inhibition may lead to adrenergic hypersensitivity from peripheral denervation and/or sympathetic denervation.

Thus, in CRPS I, the early inhibition of central cutaneous vasoconstrictor activity leads to vasodilation in the denervated area causing it to feel warm. The later increased sensitivity to circulating catecholamines due to upregulation of cutaneous adrenoreceptors causes vasoconstriction and coolness. Interestingly, studies of direct nerve injuries (CRPS II) show the same results. Initially, vasodilation is present within the denervated area, causing the skin adjacent and on the same side to become abnormally warm at first and then change to a chronically cold status. Other mechanisms include an increased density of cutaneous α-adrenoreceptors and a pathological upregulation of α-adrenergic receptors. [14, 1, 31, 25, 33, 34, 35, 36, 27]

Based on recent clinical studies, patients with neuropathic pain presenting with similar clinical signs and symptoms can be clearly divided into 2 groups by the positive and negative effects of selective sympathetic blockade, selective activation of sympathetic activity, and antagonism of α-adrenergic receptor mechanisms. [37] Pain relieved by sympatholytic procedures is considered to be SMP. SMP is now defined as a symptom or underlying mechanism in a subset of patients with neuropathic disorders. CRPS is one such neuropathic disorder. However, SMP is not a clinical entity per se. Nor is it a sine qua non for CRPS as was previously believed. Thus, the positive effect of sympathetic blockade is not essential for the diagnosis of CRPS. On the other hand, the only way to differentiate between SMP and sympathetically independent pain (SIP) is to test the efficacy of a correctly applied sympatholytic intervention. [38]

Sensory and motor dysfunction

In both types of CRPS, peripheral and central sensitization explain the pathophysiology of spontaneous pain and hyperalgesia. [39] Clinical findings in patients consistently show sensory impairments that spread beyond the injured territory, and spontaneous pain that often engulfs a quadrant or hemisensory region. These abnormal patterns are due to altered central afferent processing and have been delineated with functional imaging studies. [40, 41, 42]

Likewise, the evidence to date supports the presence of similar mechanisms involving abnormalities of CNS motor processing (rather than pain, edema, disuse, trophic changes, or nerve injury) that are responsible for causing impairments of muscle strength in the involved distal extremity. Kinematic analysis studies suggest that motor deficits are probably due to impaired integration of visual and sensory afferent input within the parietal cortex. [43] Also, an increased amplitude of physiological tremor due to CNS mechanisms is common, occurring in about 50% of patients under observation. [44]

CRPS can also be linked to structural and functional changes in the brain cortex related to sensory and motor function. Patients with early-onset CRPS demonstrate decreased cerebral perfusion and grey matter volume in the somatosensory cortex. Patients with late-onset disease demonstrate decreased cerebral perfusion in the motor cortex. [45] These findings indicate CRPS has effects on higher level motor and sensory processing in the CNS.

Aberrant healing and exaggerated inflammation

After tissue injury, the body's response is programmed to promote healing, with the goal of regaining full use of the injured body part. Some experts have hypothesized that CRPS is caused by an aberrant healing response that includes exaggerated and persistent inflammation and guarding.

At the site of injury, peripheral C-fiber nociceptors transmit pain messages that cause ortho- and retrograde release of SP and CGRP into the damaged tissues, resulting in vasodilation, extravasation of pronociceptive mediators, reactivation and further sensitization of C-fiber afferents, and increased tissue comorbidity in the injured area. [14, 10] These neuropeptides prompt the physical signs of inflammation, including redness, warmth, and swelling, that are also commonly present in early CRPS. Also, algogenic substances are released, which increase nociception and initiate the process of peripheral sensitization previously discussed. Skin sensitivity and tenderness spread into adjacent regions, which are thought to be caused by secondary hyperalgesia from CNS alterations that are consistent with the described sensitization process.

Protective disuse

Decreased use of an injured body part would appear to be a normal postinjury reaction. After injury, the organism protects and guards the injured body part to optimize healing and prevent reinjury. A normally healing organism gradually increases its use of the injured region, which aids in recovery and reintegration of the body part into the organism’s normal sense of self. However, excessive protection and guarding, such as casting or splinting, is commonly promoted by care providers, increasing the patient’s disuse of the extremity and promoting fear-avoidance, which may progress into a neurological neglect-like syndrome.

This phenomenon has been postulated as a cause in some patients with CRPS. [44] Many of the symptoms and signs of CRPS are consistent with those that would naturally develop from lack of use. For example, an unused dependent limb eventually develops swelling (dependent edema), coolness (decreased blood flow), and trophic changes (decreased blood flow). [14, 1, 46]




A population-based study by Sandroni et al showed an incidence of approximately 5.5 per 100,000 person-years at risk and a prevalence of about 21 per 100,000 for CRPS type I. [47] The same study showed an incidence of 0.8 per 100,000 and a prevalence of about 4 per 100,000 for CRPS type II. [47, 14] Therefore, the incidence of CRPS type I is higher than that of CRPS type II. [47, 14] The reported incidence of CRPS type I is 1-2% after various fractures [14] , while that of CRPS type II approximates 1-5% after peripheral nerve injury [14, 48] . The incidence of CRPS is 12% after a brain injury [49] and 5% after a myocardial infarction [50] .

A study from the Netherlands showed a total incidence of CRPS of approximately four times higher than the incidence rate observed in the only other population-based study, performed in Olmsted County, USA. [51] The estimated overall incidence rate of CRPS was 26.2 per 100,000 person years with females affected at least three times more often than males. The highest incidence occurred in females aged 61-70 years. The upper extremity was affected more frequently than the lower extremity and a fracture was the most common precipitating event (44%). [51]


Despite treatment, many patients are left with varying degrees of chronic pain, trophic changes, and disability. Pain is the most important factor leading to disability. Some have suggested that the aggressive treatment of pain in an acute setting could reduce the incidence of CRPS type I; however, further studies are needed to support this contention. Remissions followed by relapses have also been described. The frequency of the HLA-DQ1 antigen appears to be higher in patients with CRPS than in controls, and HLA-DR13 is associated with progression towards multifocal or generalized dystonia. [52, 53] Recently, a new HLA I locus was detected that may predict the spontaneous onset of CRPS. [54]

Race-, sex-, and age-related demographics

CRPS affects all races; no differences in incidence or prevalence have been observed.

Females experience CRPS more commonly than males do by a ratio that varies from 2:1 to 4:1. [14, 55, 47, 56, 57, 4, 58]

CRPS is distributed across age groups, but reaches its peak incidence between 37 and 50 years. [14, 55, 47, 56, 57, 4, 58] CRPS has an increased incidence in adolescents, compared with children, with females affected more frequently at a ratio of 4:1 and increased occurrence in the lower extremities rather than the upper by a ratio of 5.3:1. The mean age of onset is 12.5 years in a cohort of 396 children. The highest incidence of the disease appears to be in adults aged 40-49 years. CRPS appears frequently in almost every age group except children. CRPS type I has been seen in children, but the incidence is much lower than in adults.