Screening and Diagnostic Tools for Complex Regional Pain Syndrome

A Systematic Review

Giulia Mesarolia; Amos Hundert; Kathryn A. Birnie; Fiona Campbell; Jennifer Stinson


Pain. 2021;162(5):1295-1304. 

In This Article


Study and Tool Selection

The primary search to identify and summarize the measurement properties of diagnostic tools for CRPS identified 20 studies involving 4 diagnostic tools for inclusion in the review. The search of electronic databases provided a total of 6444 citations. After eliminating duplicates, reviewing abstracts, full text of the remaining studies (n = 35) were reviewed in detail. From this review, 15 did not meet the inclusion criteria, resulting in 20 included studies. Refer to Figure 1A for the PRISMA flow diagram.[25]

Figure 1.

(A) Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for primary search. (B) Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for secondary (pediatric only) search.

The secondary search to summarize diagnostic tools for CRPS used in pediatric populations identified a total of 64 studies involving 10 diagnostic tools. The search of electronic databases provided a total of 831 citations; after eliminating duplicates and reviewing abstracts, the full text of remaining studies (n = 174) were reviewed in detail. From this review, 107 did not meet inclusion criteria, resulting in 67 included studies. Please refer to Figure 1B, PRISMA flow diagram for full details.[25]

Characteristics of Included Tools

From our primary search we identified 13 unique diagnostic tools, but no screening tools. Four diagnostic tools were evaluated in more than one article by different investigatory groups: the Veldman criteria, IASP criteria, Budapest Research Criteria, and Budapest Criteria. Table 1 summarizes 9 diagnostic tools that were excluded as the measurement properties were only evaluated in one peer-reviewed article, organized in order of publication year. Findings related to these 4 tools are reported below in order of date of first publication. Table 2 summarizes the criteria included in each of these 4 tools, and Table 3 summarizes the measurement properties of each tool per each of the 20 studies included in this review. Specific details on validity (particularly specificity, sensitivity, and predictive validity) and reliability are included in Table 3. Sensitivity refers to the proportion of patients with the disease (CRPS) who test positive, and specificity refers to the proportion of patients without the disease who test negative.[9] Predictive validity refers to the probability of the disease, given the test results, expressed in positive (probability of disease in patients who test positive) and negative (probability of absence of disease in patients who test negative) predictive values.[9]

There were no studies identified that evaluated the measurement properties of the 4 diagnostic tools in a well-defined cohort of children or youth (up to 21 years of age). Mean age of study participants for each included study is reported in Table 3. Most studies did not report the age range; therefore, it remains unclear whether any children or youth were included in these samples.[4,5,11,12,14,28,29,34–36,39,40] Three studies reported the age range with the lower limit > 21 years.[17,44,47] Two studies reported the age range including lower limits < 21 years (Ott:[30] range 10–85, mean 50.9; and Yim:[46] range 16–72, mean 40.5). However, these 2 studies did not describe how many participants were younger than 21 years of age, and given the mean age reported, it is likely a small proportion. Furthermore, the data analysis was performed on the entire set of participants and therefore it is difficult to make an inference on the diagnostic validity in participants younger than 21 years of age.

Veldman Criteria. Veldman criteria were first published in 1993 to diagnose reflex sympathetic dystrophy.[43] According to these criteria, a diagnosis can be made if: (1) 4 of 5 are present (unexplained diffuse pain, difference in skin color relative to the other limb, diffuse edema, difference in skin temperature relative to the other limb, or limited active range of motion); (2) occurrence or increase of above signs and symptoms after use; and (3) above signs and symptoms present in an area larger than the area of primary injury/operation and including the area distal to the primary injury.[43] These criteria do not require any specific tools or equipment. The original criteria published by Veldman in 1993[43] use the terms "signs and symptoms" in criteria 2 and 3, and subsequent versions of the criteria published in the studies[28,34–36] that evaluate the criteria only use the term "symptoms" in criteria 1 to 3. For example, related to skin color changes, it is unclear if a patient would meet the criteria if they reported skin color changes at any time (symptom), or if the physician must observe this at the time of examination (sign).

Six studies evaluated the measurement properties of the Veldman criteria,[28,30,33–36] 4 of which were from the same investigatory group. Five studies were exclusive to CRPS Type 1 in the Dutch population. Perez[33] evaluated the discriminant validity of the individual criteria. The study by Ott et al.[30] was the only study to report the tool's overall sensitivity (67%) and specificity (87%). Two studies evaluated the construct validity of the individual items by comparing physician examination to objective measures of symptoms (edema, temperature, range of motion, and pain).[28,33] For example, physicians' assessment of edema as compared to volumetric measurement. Both studies reported good agreement (51%-96% agreement) between physician examination and objective measures across symptoms with respect to presence or absence of symptoms,[28,33] but poor correlations (0%-71% association) with respect to symptom severity.[33] Similar results were found with respect to interrater reliability in another study, which reported that physicians agreed upon the presence or absence of symptoms (range 88%-100%) but poor agreement with respect to symptom severity, particularly with temperature and discoloration.[34] With respect to concurrent validity, its relatedness to the Budapest Criteria and Budapest Research Criteria was quite poor (Cohen's kappa coefficient [κ] range 0.29–0.42), resulting in disagreement in diagnosis between the tools in 26% to 39% of study participants.[35]

International Association for the Study of Pain Criteria. The IASP criteria for diagnosing CRPS were first created at an IASP meeting in Orlando in 1993, and later published in 1994.[21] According to the IASP criteria, a patient is diagnosed with CRPS if they meet all 4 criteria: (1) presence of an initiating noxious event or a cause of immobilization; (2) continuing pain, allodynia, or hyperalgesia for which the pain is disproportionate to any inciting event; (3) evidence at some time of edema, changes in skin blood flow, or abnormal sudomotor activity in the painful region; and (4) diagnosis is precluded by the existence of conditions that would otherwise account for the degree of pain and dysfunction. Our search identified 10 studies that evaluated the measurement properties of the IASP criteria.[4,11–14,29,30,35,40,44] Many studies examined the individual criteria in the tool, including the sensitivity, specificity,[11,35] and predictive validity[11,13] of each sign and symptom. The overall sensitivity and specificity of the tool ranges from 85% to 100% and 36% to 60%, respectively.[4,13,29,30] Oh et al.[29] examined the diagnostic validity specifically in poststroke patients (sensitivity 100% and specificity 41%). Galer et al.[11] found that the positive predictive value and specificity of individual criteria were poor overall, although the specificity was greater for signs as opposed to symptoms. One-third of patients with diabetic neuropathy met the IASP criteria for diagnosis.[11] With respect to concurrent validity and its relatedness to other tools, the IASP criteria were found to have a large association with the CRPS Severity Score (Eta = 0.69, where Eta represents a nonlinear correlation coefficient with a range from 0 to 1.00).[14] Interrater reliability was examined, and the Cohen's Kappa (κ) value was 0.29 (CI: 0.03–0.55), which was preferable compared with physician diagnosis (κ = 0.20) but not as strong as the Budapest Research Criteria (κ = 0.38).[44]

Harden et al.[12] conducted a factor analysis of the individual criteria. They suggested further modification of the diagnostic criteria, specifically to separate edema, vasomotor, and sudomotor symptoms into distinct categories rather than combined as one. The most significant factors included sensory changes, temperature asymmetry, color changes, edema, sweating, and motor dysfunction. From these results, Harden et al.[12] proposed a new set of criteria, which were later named the Budapest Research Criteria.

Budapest Research Criteria. These criteria were first introduced by Bruehl et al.[4] as a modified version of the IASP criteria, intended for use in research studies to define study populations. The Bruehl et al.[4] criteria have 3 components for diagnosis: (1) continuing pain, which is disproportionate to any inciting event; (2) one symptom in each of 4 categories; and (3) one sign in 2 of 4 categories; categories include sensory, vasomotor, sudomotor/edema, and motor/trophic. These criteria were later revised to include an additional fourth one, named the Budapest Research Criteria.[8] The added criteria stipulate that no other diagnosis could better explain the patient's presentation. The Bruehl et al.[4] criteria were evaluated in 7 research studies.,[4,17,30,35,41,42,44] and the Budapest Research Criteria in 5 studies.[13,14,40,46,47] Harden et al.[13] reported that the Budapest Research Criteria were found to have a more balanced profile of sensitivity (78%) and specificity (79%) compared with the IASP or Budapest Criteria. Similar results were found even when the cutoff scores were modified.[46] By contrast, Ott[30] and Sumitani[40] report the opposite, both noting a more polarized profile with poor sensitivity (20%-41%) and excellent specificity (94%-95%). With respect to concurrent validity, the Budapest Research Criteria had a high degree of relatedness to the Atkins diagnostic criteria (κ = 0.79) and the CRPS Severity Score (Eta = 0.77). Interrater reliability was found to be moderate (κ = 0.38). The discriminant validity of individual signs and symptoms was examined,[35] with another notable study[17] that examined the ability of temperature differences to discriminate between CRPS, healthy control, and people with other types of limb pain. Krumova[17] concluded that a temperature side difference of 2°C resulted in a sensitivity of 73% and specificity of 94%.

Budapest Criteria. In 2003, the IASP held a consensus conference in Budapest with a view to improve the IASP diagnostic criteria for CRPS. This meeting resulted in the creation of the new criteria, named the Budapest Criteria.[15] These criteria largely reflected criteria proposed earlier by Harden and Bruehl in 1999.[4,12] The Budapest Criteria includes motor and trophic features of the disease and more emphasis on signs (ie, observable by the clinician). The Budapest Criteria, intended for clinical purposes, mirror the Budapest Research Criteria with the exception of a difference in scoring. Budapest Criteria are (1) continuing pain that is disproportionate to any inciting event, (2) one symptom in 3 of 4 categories, and (3) one sign in 2 of 4 categories; categories include sensory, vasomotor, sudomotor/edema, and motor/trophic, and (4) no other diagnosis could better explain the patient's presentation. These criteria underwent initial validation in 2010 by Harden et al.,[13] where the Budapest Criteria were compared to the IASP criteria in discriminating between CRPS-1 and other types of neuropathic pain. This study concluded that both criteria had excellent diagnostic sensitivity (IASP criteria 100% and Budapest Criteria 99%), but the Budapest Criteria had superior specificity (68%) compared with the IASP criteria (41%).[13] Analysis of the discriminant validity of individual criteria was also performed in this study, with sensitivity (93%-94%) and specificity (57%-71%). Overall, the tool's diagnostic validity has been evaluated in 4 studies with highly variable results, particularly with respect to the sensitivity (sensitivity ranging from 45% to 99%) and, to a lesser extent, specificity (68%-85%).[13,29,30,40] Three additional studies evaluated additional measurement properties of the Budapest Criteria.[14,19,40] Mailis-Gagnon compared the Budapest Criteria with clinical diagnosis and found that only 27% of patients diagnosed by a community-based provider met the Budapest Criteria.[19] This study reported that more than 80% of patients who did not meet the criteria had another diagnosis to better explain their signs and symptoms, which is the fourth Budapest Criteria.[19] Concurrent validity with the CRPS Severity Score was excellent (Eta = 0.88).

Quality of Studies Examining the 4 Diagnostic Tools. QUADAS-2 risk-of-bias ratings for each study are presented in Table 4. Figure 2 summarizes overall risk of bias and applicability concerns. Overall, the majority of studies showed low concern regarding applicability for the reference standard, index test, and patient selection. Scores for risk of bias indicated greater concern with more than half of studies demonstrating high or unclear risk of bias concerns for the flow and timing, reference standard, index test, and patient selection.

Figure 2.

Summary of study risk of bias and applicability concerns, QUADAS-2.

Diagnostic Tools Used in Pediatric Studies

Our secondary search identified 67 studies examining pediatric CRPS and extracted data on the study type as well as the tools used to diagnose CRPS in their study population. Table 5 summarizes findings from this review and lists 10 diagnostic tools that were reportedly used in pediatric studies of CRPS. More than half of the studies identified in this review used no specific criteria to diagnose CRPS in the study population. Ten diagnostic tools were used in total, 4 of which were previously established tools (the Budapest Criteria, IASP criteria, Veldman criteria, and Japanese Diagnostic Criteria). Six studies used a unique set of diagnostic criteria defined by the study authors to diagnose CRPS. No studies used the Budapest Research Criteria to define their study population. Of note, a large proportion (37%) of studies identified were case studies or series, and only 21% of studies were interventional.