`
`External validation of IASP diagnostic criteria for Complex Regional Pain
`Syndrome and proposed research diagnostic criteria
`
`Stephen Bruehl a,*, R. Norman Hardena, Bradley S. Galer b, Samuel Saltz a, Martin Bertram c,
`Miroslav Backonjad, Richard Gaylese, Nathan Rudinf,
`Maninder K. Bhugrag, Michael Stanton-Hicks g
`aCenter for Pain Studies, 11th Floor, Rehabilitation Institute of Chicago, 345 E. Superior St., Chicago, IL 60611, USA
`bBeth Israel Medical Center, Pain Service, New York, NY, USA
`cPain Management Clinic, Wright Patterson Air Force Base, Cleveland, OH, USA
`dDepartment of Neurology, University of Wisconsin-Madison, Madison, WI, USA
`eSpace Coast Anesthesiology, Cocoa Beach, FL, USA
`fDepartment of Physical Medicine and Rehabilitation, Johns Hopkins University School of Medicine, Baltimore, MD, USA
`gPain Management Center, Cleveland Clinic Foundation, Cleveland, OH, USA
`
`Received 22 July 1998; received in revised form 17 November 1998; accepted 24 December 1998
`
`Abstract
`
`Recent work in our research consortium has raised internal validity concerns regarding the current IASP criteria for Complex Regional
`Pain Syndrome (CRPS), suggesting problems with inadequate sensitivity and specificity. The current study explored the external validity of
`these IASP criteria for CRPS. A standardized evaluation of signs and symptoms of CRPS was conducted by study physicians in 117 patients
`meeting IASP criteria for CRPS, and 43 patients experiencing neuropathic pain with established non-CRPS etiology (e.g. diabetic neuro-
`pathy, post-herpetic neuralgia). Multiple discriminant function analyses were used to test the ability of the IASP diagnostic criteria and
`decision rules, as well as proposed research modifications of these criteria, to discriminate between CRPS patients and those experiencing
`non-CRPS neuropathic pain. Current IASP criteria and decision rules (e.g. signs or symptoms of edema, or color changes or sweating
`changes satisfy criterion 3) discriminated significantly between groups (P , 0.001). However, although sensitivity was quite high (0.98),
`specificity was poor (0.36), and a positive diagnosis of CRPS was likely to be correct in as few as 40% of cases. Empirically-based research
`modifications to the criteria, which are more comprehensive and require presence of signs and symptoms, were also tested. These modified
`criteria were also able to discriminate significantly, between the CRPS and non-CRPS groups (P , 0.001). A decision rule, requiring at
`least two sign categories and four symptom categories to be positive optimized diagnostic efficiency, with a diagnosis of CRPS likely to be
`accurate in up to 84% of cases, and a diagnosis of non-CRPS neuropathic pain likely to be accurate in up to 88% of cases. These results
`indicate that the current IASP criteria for CRPS have inadequate specificity and are likely to lead to overdiagnosis. Proposed modifications
`to these criteria substantially improve their external validity and merit further evaluation.
`1999 International Association for the Study
`of Pain. Published by Elsevier Science B.V.
`
`Keywords: Complex regional pain syndrome; Reflex sympathetic dystrophy; Causalgia; Validation; Diagnosis; Diagnostic criteria
`
`1. Introduction
`
`The publication in 1994 of standardized, consensus-based
`diagnostic criteria for Complex Regional Pain Syndrome
`(CRPS) was a step forward in the diagnosis of regional
`
`* Corresponding author. Tel.: +1-312-908-9501; fax: +1-312-908-1833.
`
`pain disorders associated with vasomotor or sudomotor
`changes (Merskey and Bogduk, 1994). This syndrome was
`known previously by various names, most commonly reflex
`sympathetic dystrophy and causalgia, and was diagnosed
`using a variety of non-standardized or incompatible diag-
`nostic schemes (Kozin et al., 1981; Amadio et al., 1991;
`Blumberg, 1991; Gibbons and Wilson, 1992). The standar-
`dized CRPS criteria published by the International Associa-
`
`0304-3959/99/$20.00 (cid:211)
`1999 International Association for the Study of Pain. Published by Elsevier Science B.V.
`PII: S0304-3959(99)00011-1
`
`Grun Exh. 1046
`PGR for U.S. Patent No. 9,283,239
`
`
`1
`
`Grün. Exh. 1032
`PGR for U.S. Patent No. 10,052,338
`
`(cid:211)
`
`
`148
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`tion for the Study of Pain (IASP; Merskey and Bogduk,
`1994) are intended to improve clinical recognition of the
`disorder, and facilitate selection of more generalizable sam-
`ples for treatment outcome and basic science research (Stan-
`ton-Hicks et al., 1995; Wilson et al., 1996).
`Widespread use of these standardized CRPS diagnostic
`criteria has the potential to lead to improved understanding
`and treatment of the disorder. However, realization of this
`potential is limited by the fact that the criteria were derived
`rationally, based upon the consensus of a group of expert
`clinicians. While this was an appropriate first step towards
`developing criteria, experience regarding criterion develop-
`ment in the areas of headache and psychiatric diagnosis
`highlight the necessity of validating, and if necessary mod-
`ifying, these initial consensus-based criteria based upon
`results of validation research and further clinical experience
`(Merikangas and Frances, 1993). Although the IASP criteria
`were published nearly 4 years ago, research to validate them
`empirically has been quite limited. In the absence of ade-
`quate research, the validity of these criteria remains uncer-
`tain, and the possibility of significant under- or over-
`diagnosis cannot be excluded (Galer et al., 1998).
`The limited available work in this area suggests several
`problems with the current IASP criteria. One validity issue
`is internal validity, which addresses the extent to which
`interrelationships between CRPS signs and symptoms
`observed in clinical patients correspond to the IASP criteria.
`Recent work by the current authors used principal compo-
`nents factor analysis (PCA) to study the internal validity of
`the IASP criteria (Bruehl et al., 1998). PCA is a statistical
`technique which identifies coherent, and presumably con-
`ceptually-linked, subsets of variables within a dataset.
`Unlike the IASP diagnostic scheme which treats edema,
`vasomotor, and sudomotor changes as a unitary criterion,
`PCA indicated that signs and symptoms of vasomotor
`change form a factor which is statistically-distinct from
`sudomotor changes and edema (which did group together).
`The fact that these two statistically-distinct groups of signs
`and symptoms are combined into a single criterion in the
`IASP criteria, may contribute to their poor specificity
`(Bruehl et al., 1998). In addition to the findings above,
`PCA revealed the presence of statistically-distinct compo-
`nents of CRPS which are not incorporated in the current
`criteria: weakness, movement disorder, tremor, dystonia,
`diminished range of motion and trophic changes of the
`hair, skin and nails (Bruehl et al., 1998). These signs and
`symptoms form a distinct cluster which is referred to, here-
`after, as a motor/trophic cluster.
`External validity of the IASP criteria is another important
`issue to be addressed. External validity of diagnostic criteria
`refers to their usefulness for distinguishing between patients
`on the basis of some external reference or ‘gold standard’
`(Merikangas et al., 1994). The external validity of the IASP
`criteria was recently examined in a small pilot study by
`members of our CRPS research consortium (Galer et al.,
`1998). This study examined the ability of the IASP diag-
`
`nostic criteria to distinguish between CRPS and diabetic
`neuropathy patients. Use of current IASP criteria and deci-
`sion rules (e.g. criterion 3 is satisfied by presence of edema
`or skin blood flow changes or sweating changes) to make
`diagnostic decisions led to substantial overdiagnosis. If
`objective test results to identify diabetes were unavailable
`and diagnosis were made solely on the pattern of signs and
`symptoms, up to 37% of diabetic neuropathy patients were
`likely to be misdiagnosed as having CRPS (Galer et al.,
`1998). Although based on a small sample, results of this
`study also suggested that modification of the IASP criteria
`and decision rules might substantially improve their diag-
`nostic accuracy.
`The current study was designed to provide a more com-
`prehensive evaluation of the external validity of the IASP
`CRPS criteria. As in the Galer et al. (1998) study, the cur-
`rent study was based upon the premise that if the IASP
`criteria and decision rules for diagnosing CRPS cannot dis-
`criminate between it and neuropathic pain disorders which
`do not have a substantial autonomic component, these cri-
`teria are likely to be of limited use clinically or for defining
`research samples. This study also sought to evaluate pro-
`posed empirically-derived research modifications to CRPS
`criteria previously suggested by our research consortium
`(Bruehl et al., 1998).
`
`2. Methods
`
`2.1. Design
`
`The study is a multi-site between-subjects design com-
`paring CRPS to non-CRPS neuropathic pain patients.
`
`2.2. Participants
`
`Participants included a series of 117 CRPS patients and
`43 patients diagnosed with non-CRPS neuropathic pain
`(non-CRPS) who presented for evaluation and treatment
`at the data collection sites. CRPS was diagnosed in all
`patients according to published IASP criteria (see Appendix
`A; Merskey and Bogduk, 1994). Objective tests of nerve
`dysfunction (EMG/nerve conduction) were available in a
`subset of CRPS patients, which if used as a diagnostic cri-
`terion, would have led to the diagnosis of CRPS-Type I in
`approximately two-thirds of the sample (Merskey and Bog-
`duk, 1994; Baron et al., 1996). Comparison of known Type I
`and Type II CRPS patients (based on absence or presence of
`objective EMG/nerve conduction abnormalities, respec-
`tively) revealed no differences in frequency of any sign or
`symptom between diagnostic groups (all P . 0.10), and
`therefore, the remaining analyses did not separate these
`diagnostic subcategories. The non-CRPS group reflected
`several known non-CRPS diagnoses including diabetic neu-
`ropathy (44.2%), polyneuropathy (14.0%), post-herpetic
`neuralgia (20.9%) and radiculopathy (20.9%). To avoid
`
`2
`
`
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`149
`
`confounding the two groups, the non-CRPS patients were
`not identified by process of exclusion (i.e. simply failing to
`meet CRPS criteria). Rather, each of the non-CRPS disor-
`ders was diagnosed using criteria distinct from CRPS cri-
`teria, such as extremity pain coexisting with known diabetes
`mellitus, or pain in a radicular pattern with disk herniation
`confirmed by MRI.
`
`2.3. CRPS database checklist
`
`In order to insure standardized assessment of signs and
`symptoms across sites, a database checklist was used. This
`CRPS checklist presents a complete list of the signs and
`symptoms used to diagnose CRPS, as well as other signs/
`symptoms (e.g.
`trophic changes, motor abnormalities)
`which are reported to be associated with the disorder in
`previous literature, but are not incorporated in the IASP
`diagnostic criteria (Schwartzman and McLellan, 1987;
`Stanton-Hicks et al., 1990, 1995; Merskey and Bogduk,
`1994; Janig and Stanton-Hicks, 1996; Wilson et al., 1996).
`As recommended by Janig et al. (1991), dichotomous mea-
`sures (i.e. presence or absence) were used to assess signs
`and symptoms due to the potential for inter-rater unreliabil-
`ity using interval rating scales. Standardized procedures for
`evaluating the different signs are provided with the checklist
`to maximize uniform assessment across sites. Signs and
`symptoms in the checklist are summarized in Appendix B.
`
`2.4. Procedures
`
`For all patients in both groups, an evaluation of signs and
`symptoms was conducted by a study physician using the
`CRPS checklist described above. This involved obtaining
`a patient history to assess symptoms, as well as conducting a
`physical examination to assess signs.
`
`2.5. Statistical analysis
`
`The primary analyses tested whether CRPS and non-
`CRPS neuropathic pain can be distinguished based upon
`patterns of signs and symptoms. These analyses were
`more specifically designed to test the ability of current
`IASP criteria and decision rules to distinguish between the
`CRPS and non-CRPS groups. The differential diagnosis of
`CRPS from disorders such as diabetic neuropathy and post-
`herpetic neuralgia, as in the current study, is unlikely in
`clinical practice. However, statistical examination of the
`ability of the IASP criteria to discriminate CRPS from
`‘known’ non-CRPS disorders (i.e. without autonomic dys-
`function) is an appropriate model for testing the external
`validity of the diagnostic criteria. Similar statistical models
`have been used in the validation of diagnostic criteria for
`headache and psychiatric disorders
`(Merikangas and
`Frances, 1993; Merikangas et al., 1994).
`Due to the suspected inadequacies in current IASP CRPS
`criteria suggested by Galer et al. (1998), the current study
`
`was also used to test a set of proposed research diagnostic
`criteria for CRPS which were empirically-derived using
`principal components factor analysis (PCA; Bruehl et al.,
`1998). Results of this study are presented in detail in a
`separate manuscript currently under review. As a pattern
`recognition technique, PCA provided an empirical basis
`for determining the proper manner in which to group
`together signs and symptoms included in a set of proposed
`CRPS research diagnostic criteria (see Appendix C).
`Briefly, these research criteria require the presence of both
`signs and symptoms, each of which are divided into four
`categories: (1) sensory (2) vasomotor (3) sudomotor/edema
`and (4) motor/trophic. While PCA allowed determination of
`the proper groupings of CRPS signs/symptoms, it did not
`permit determination of the optimal decision rules (i.e. num-
`ber of signs/symptom categories which must be positive) for
`determining presence or absence of CRPS. The methodol-
`ogy of the current study was designed to address this latter
`issue, and therefore, a series of decision rules based upon
`these research criteria was tested, each differing in the num-
`ber of sign and symptom categories required to be positive
`to meet the diagnostic threshold for CRPS.
`Discriminant function analyses (DFAs) were conducted
`using IASP and proposed research criteria/decision rules to
`discriminate between the CRPS and the non-CRPS groups.
`DFA determines a discriminant score for each case, and then
`applies Bayes’ theorem to derive a general rule for classify-
`ing cases into one of two groups. Results of DFA were then
`used to derive indices of discriminative efficiency, including
`sensitivity, specificity, positive predictive power (PPP) and
`negative predictive power (NPP). Sensitivity is defined as
`true positive rate/true positive + false negative rates,
`reflecting the percentage of true positive (CRPS) cases clas-
`sified accurately. Specificity is defined by true negative rate/
`true negative + false positive rates, and reflects the propor-
`tion of true negative (non-CRPS) cases classified accurately.
`Of more importance clinically, given the need to maximize
`probability of correct diagnosis when actual disease status is
`unknown, are PPP and NPP (Landau et al., 1991). In this
`study, PPP indicates the probability that a diagnosis of
`CRPS is accurate, whereas NPP indicates the probability
`that a diagnosis of non-CRPS neuropathic pain is accurate.
`Both PPP and NPP are in part a function of the base rate
`(prevalence) for the targeted disorder (CRPS) in the popula-
`tion being examined, and these were derived as described by
`Meehl and Rosen (1955). PPP was defined as: (CRPS base
`rate ·
`true positive rate)/((CRPS base rate ·
`true positive
`rate) + (1 - CRPS base rate ·
`false positive rate)). NPP
`was defined as: ((1 - CRPS base rate) ·
`true negative
`rate)/((1 - CRPS base rate ·
`true negative rate) + (CRPS
`base rate ·
`false negative rate)). These four indicators of
`diagnostic efficiency were contrasted across different cri-
`teria and/or decision rules to determine relative accuracy
`and likely diagnostic utility of each.
`Actual sample size values varied slightly across analyses
`due to missing data. The maximum available number of
`
`3
`
`
`
`150
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`subjects was used for all analyses. All probability values are
`two-tailed.
`
`3. Results
`
`3.1. Demographics and pain characteristics
`
`A comparison of the demographics and pain characteris-
`tics across the CRPS and non-CRPS groups is presented in
`Table 1. The two subsamples differed significantly on sev-
`eral variables. CRPS patients were younger (t (131) = 9.50,
`P , 0.001), had shorter pain duration (t (149) = 5.18, P ,
`0.001), and were more likely than non-CRPS patients to be
`experiencing upper extremity (Phi (146) = 0.54, P , 0.001)
`and/or unilateral pain (Phi (147) = 0.67, P , 0.001).
`
`3.2. Discriminant function analyses (DFA)
`
`Current IASP criteria (see Appendix A) for CRPS were
`examined first. IASP criterion 2 (continuing pain, allodynia,
`or hyperalgesia with pain disproportionate to the inciting
`event) and criterion 3 (edema, changes in surface blood
`flow, or abnormal sudomotor activity) were combined in a
`DFA to distinguish between CRPS and non-CRPS groups.
`A literal interpretation of current IASP criteria as written
`(‘evidence at some point of…’), which allows criteria to be
`met by presence of current objective signs or historical
`symptom reports, distinguished significantly between the
`two groups (chi-square (2) = 40.9, P , 0.001). Requiring
`
`Table 1
`Demographics and pain characteristics across diagnostic groups
`
`Variable
`
`Gender
`
`Race
`African–American
`Caucasian
`Hispanic
`Other
`
`Age**
`Pain duration
`
`Non-CRPS neuro-
`pathic (n = 43)
`
`CRPS
`(n = 117)
`
`50.0% Female
`
`62.6% Female
`
`13.9%
`83.3%
`2.8%
`0.0%
`
`4.6%
`88.1%
`2.8%
`4.5%
`
`61.5 (SE = 2.2)
`70.0 (SE = 14.1)
`
`41.0 (SE = 1.0)
`22.9 (SE = 2.1)
`
`Pain Location**
`Upper extremity
`Lower extremity
`Upper and lower extremity
`Other
`
`Affected Side**
`Left
`Right
`Bilateral
`
`7.9%
`60.5%
`13.2%
`18.4%
`
`10.5%
`23.7%
`65.8%
`
`CRPS, complex regional pain syndrome.
`*P , 0.05; **P , 0.01.
`
`47.4%
`51.8%
`0.0%
`0.9%
`
`49.5%
`45.9%
`4.5%
`
`Table 2
`Sensitivity and specificity of diagnostic criteria/decision rules to discrimi-
`nate CRPS from non-CRPS neuropathic conditions. Numbers listed in the
`decision rules refer to number of sign and symptom categories (out of four
`possible categories for each) required to be present for the syndrome to be
`considered CRPS
`
`Criteria/decision rule
`‡ 1 sign or symptom for both
`IASP:
`criterion 2 and 3*
`‡ 1 sign for both criterion
`IASP:
`2 and 3*
`‡ 2 sign categories
`Research criteria:
`and ‡ 2 symptom categories*
`‡ 2 sign categories
`Research criteria:
`and ‡ 3 symptom categories*
`‡ 2 sign categories
`Research criteria:
`and 4 symptom categories*
`‡ 3 sign categories
`Research criteria:
`and ‡ 2 symptom categories*
`‡ 3 sign categories
`Research criteria:
`and ‡ 3 symptom categories*
`‡ 3 sign categories
`Research criteria:
`and 4 symptom categories*
`
`*P , 0.001.
`
`Sensitivity
`
`Specificity
`
`0.98
`
`0.82
`
`0.94
`
`0.85
`
`0.70
`
`0.76
`
`0.70
`
`0.86
`
`0.36
`
`0.60
`
`0.36
`
`0.69
`
`0.94
`
`0.81
`
`0.83
`
`0.75
`
`the presence of objective signs (strict interpretation) for a
`diagnosis of CRPS to be made also resulted in significant
`(chi-square (2) = 23.9,
`discrimination between groups
`P , 0.001).
`Proposed research criteria (Appendix C) were also tested
`using a variety of decision rules for determining the thresh-
`old for diagnosis of CRPS. Decision rules ranging from a
`requirement that at least two of four sign categories and at
`least two of four symptom categories be positive (chi-square
`(2) = 18.8, P , 0.0010), to a more stringent rule that three
`+ sign categories and four symptom categories be positive
`(chi-square (2) = 68.1, P , 0.001), all discriminated sig-
`nificantly between the CRPS and non-CRPS groups (see
`Table 2 for complete list of decision rules tested).
`
`4. Diagnostic efficiency
`
`Although it might be assumed that the various diagnostic
`criteria and decision rules tested are roughly comparable,
`given that all DFAs were significant, examination of data
`regarding diagnostic efficiency of each revealed substantial
`differences. Table 2 presents sensitivity and specificity for
`all criteria and decision rules tested. The literal interpreta-
`tion of current IASP criteria (satisfied by presence of signs
`or symptoms) resulted in a high level of sensitivity, but
`quite poor specificity. A strict interpretation of IASP cri-
`teria (must display objective signs) was also associated
`with good sensitivity, but only moderately improved spe-
`cificity. Results for both interpretations of IASP criteria
`were consistent with results of similar analyses reported
`by Galer et al. (1998) which found specificities of 0.27
`
`4
`
`
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`151
`
`Fig. 1. (A) Positive predictive power (PPP) and negative predictive power (NPP) for proposed research diagnostic criteria with 2 + sign and 3 + symptom
`decision rule. The x-axis reveals how PPP and NPP vary across all hypothetical base rates (prevalence) for CRPS within the larger population of neuropathic-
`type disorders seen in tertiary pain management centers. PPP and NPP values refer to probability that a diagnosis of CRPS is accurate (PPP) or a diagnosis of
`non-CRPS pain is accurate (NPP). (B) Two + sign and four symptom decision rule. (C) Three + sign and two + symptom decision rule. (D) Three + sign and
`three + symptom decision rule. (E) Three + sign and four symptom decision rule.
`
`and 0.55 for the literal and strict interpretations of IASP
`criteria.
`As noted above, positive (PPP) and negative (NPP) pre-
`dictive power may be of more relevance clinically, given
`that they directly reflect the probability that a given diag-
`nosis of CRPS or Non-CRPS pain, respectively, is correct.
`Fig. 1A–E display PPP and NPP for the five decision rules
`which displayed the best combinations of sensitivity and
`specificity in Table 2. Since the actual PPP and NPP of a
`given diagnostic decision rule is in part a function of the
`prevalence of CRPS in the pain population of interest
`(Meehl and Rosen, 1955), each of these figures displays
`PPP and NPP for all possible CRPS prevalence rates. For
`
`example, in Fig. 1B, the PPP line represents the probability
`of an accurate CRPS diagnosis across all CRPS prevalence
`rates from 0 to 100%. Assuming that CRPS occurs in
`approximately 25% of patients with unexplained neuro-
`pathic pain who are sent to tertiary care clinics, this graph
`indicates that using the modified research criteria, requiring
`two or more sign categories and four symptom categories to
`be positive is likely to result in accurate diagnosis of CRPS
`in 80% of cases. With this same 25% prevalence rate for
`CRPS, likely accuracy of a non-CRPS diagnosis (the NPP
`line) is approximately 90%. Examination of Fig. 1A–E
`clearly indicates that a decision rule requiring that two or
`more sign categories and four symptom categories be posi-
`
`5
`
`
`
`152
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`tive would provide the highest combination of PPP and
`NPP across the widest range of possible CRPS prevalence
`rates.
`
`5. Discussion
`
`The IASP criteria for CRPS were developed by a group of
`clinicians and basic scientists during a Dahlem-type work-
`shop (Stanton-Hicks et al., 1995). The resulting criteria
`were intended to facilitate the diagnosis of CRPS and pro-
`mote research through the use of a descriptive system, rather
`than one relying on assumptions about pathophysiology
`(Stanton-Hicks et al., 1995). The authors of the new taxon-
`omy expressed the goal that uniform criteria would improve
`the clinical recognition of the disorder, facilitate more gen-
`eralizable treatment outcome research, and lead to identifi-
`cation of possible new subcategories of
`the disorder
`(Stanton-Hicks et al., 1995). However, a Medline search
`reveals only one study to date specifically addressing the
`validity of these IASP criteria (Galer et al., 1998). The
`current study is an empirical evaluation of the IASP
`CRPS criteria and their external validity, and an effort to
`acquire data for use in guiding future revisions of these
`criteria.
`Consistent with pilot work by Galer et al. (1998), the
`results of the current study indicate that the external validity
`of the IASP criteria can be challenged; their application as
`currently written may result in the over-diagnosis of CRPS.
`A proposed set of research diagnostic criteria for CRPS,
`based upon results of the current study and previous factor
`analysis research (Bruehl et al., 1998) appears to be more
`specific than current IASP criteria, and may substantially
`improve the ability to discriminate accurately between
`CRPS and other types of neuropathic pain. Decision rules
`tested in this study based on these modified research criteria
`suggest that a rule requiring at least two of four sign cate-
`gories and four symptom categories to be positive maxi-
`mized diagnostic accuracy across the widest range of
`CRPS prevalence rates. The discriminative ability of the
`modified research criteria tested in this study suggests that
`further research regarding these criteria is merited. It is
`hoped that data from this and other similar studies may
`provide data-based guidelines for future revisions of the
`criteria by the IASP taxonomy committee.
`One issue not addressed in the current study is the possi-
`ble impact of CRPS with nerve injury (Type II) versus
`CRPS without such injury (Type I) on the diagnostic effi-
`ciency of CRPS criteria. The developers of the IASP criteria
`for CRPS did not assume that it was a unitary phenomenon,
`and this was reflected in the requirement that CRPS Type I
`versus Type II be specified when making a diagnosis. In the
`current study, this issue was not specifically examined given
`that analyses of these data had revealed no significant dif-
`ferences in frequency of various signs and symptoms
`between patients with and without nerve injury as documen-
`
`ted by EMG/NCV (Bruehl et al., 1998). This finding sug-
`gested, that it was unlikely that the presence or absence of
`nerve injury would have a significant impact on the ability
`of the IASP criteria and the proposed research criteria to
`identify CRPS accurately. However, when a larger set of
`data is available, it would be useful to test this issue by
`replicating the current study using CRPS groups with and
`without documented nerve injury.
`The differential diagnosis specifically between CRPS and
`post-herpetic neuralgia or diabetic neuropathy is likely to be
`uncommon in typical clinical situations. However, this study
`used a methodology similar to that used in other diagnostic
`validity research (Merikangas and Frances, 1993; Merikan-
`gas et al., 1994; Galer et al., 1998) which allowed a con-
`trolled test of a statistical model analogous to the clinical
`process of CRPS diagnosis. The pain physician is frequently
`presented with a patient experiencing an unidentified pain
`complaint suspected to be neuropathic, with the task of iden-
`tifying it properly and planning treatment accordingly. The
`IASP diagnostic criteria are designed to provide an objective
`means of making decisions as to whether such unidentified
`conditions are CRPS (i.e. in which autonomic dysfunction is
`present) or some other type of neuropathic pain. Treatment
`for these two types of conditions will differ, and application
`of inappropriate (and possibly expensive) treatments due to
`misdiagnosis may contribute to excessive medical costs, or
`worse, delay more appropriate treatment in some cases.
`Therefore, empirically-guided revisions which improve the
`validity of the CRPS diagnostic criteria may impact posi-
`tively on problems of medical over-utilization and patient
`quality-of-life. Such improvements to the CRPS criteria will
`also assist in identifying more appropriate research samples
`to evaluate and improve therapeutic outcomes (Stanton-
`Hicks et al., 1998).
`One potential issue regarding the methodology of this
`study is that CRPS criteria were used to define the CRPS
`group, but in some analyses were also used to discriminate
`between diagnostic groups. This procedure was used
`because there is no ‘gold standard’ (i.e. single known patho-
`physiology) for identifying CRPS independent of the cur-
`rent consensus-based IASP criteria. Although not ideal, this
`methodology is unlikely to have confounded the results for
`several reasons. First, the non-CRPS group was defined
`using criteria which were independent of the CRPS criteria,
`rather than simply reflecting failure to meet CRPS criteria.
`Selection of the non-CRPS group using this latter methodol-
`ogy would clearly have produced invalid results. Second, if
`this methodology had confounded the results, it should have
`maximized group differences, thus making it easier to dis-
`criminate statistically between groups. The fact that the
`current IASP criteria displayed poor specificity despite
`this possible exaggeration of group differences reaffirms
`the need for more thorough validation of the current criteria
`in large clinical populations. Finally, tests of the proposed
`research modifications to the IASP CRPS criteria and deci-
`sion rules did not suffer from this same potential confound.
`
`6
`
`
`
`S. Bruehl et al. / Pain 81 (1999) 147–154
`
`153
`
`Methodology similar to that described above has been
`used in the process of validating headache diagnostic cri-
`teria as well, given a similar absence of clear diagnostic
`markers for those disorders (Merikangas and Frances,
`1993; Merikangas et al., 1994). Researchers attempting to
`validate headache diagnostic criteria have accepted the fact
`that with a lack of a defined pathophysiology, and therefore,
`an absence of definitive validation studies, the emphasis
`must be on repeated evaluation of the validity of diagnostic
`criteria using the best means that are available (Merikangas
`et al., 1994). It is hoped that the current study will spur
`additional research focused on evaluation of CRPS diagnos-
`tic validity.
`Results of this study confirm the existence of a syndrome
`which is statistically distinguishable from other types of
`known neuropathic pain, and suggest modifications which
`may enhance CRPS diagnostic accuracy. Although the mod-
`ified research criteria examined in this study appear promis-
`ing and deserve further research, use of these modified
`criteria for ‘clinical’ diagnosis is clearly premature. This
`would encourage a reversion to the situation prior to 1994
`in which there was no universal standard for CRPS diagno-
`sis, and would consequently, defeat the original purpose of
`the standardized IASP criteria. The proposed modified
`research criteria should, therefore, be used only for research
`until sufficient validation data are available to justify a for-
`mal revision of the CRPS criteria through the IASP taxon-
`omy
`committee.
`Further
`exploration
`of
`potential
`modifications to the IASP diagnostic scheme for CRPS
`may be invaluable for guiding such revisions of the CRPS
`criteria, and ultimately will contribute to improved clinical
`diagnosis.
`
`Appendix A IASP diagnostic criteria for Complex
`Regional Pain Syndrome
`
`(1) The presence of an initiating noxious event, or a
`cause of immobilization.
`(2) Continuing pain, allodynia, or hyperalgesia with
`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 region
`of pain.
`(4) This diagnosis is excluded by the existence of con-
`ditions that would otherwise account for the degree of pain
`and dysfunction.
`
`Appendix B Signs and/or symptoms on CRPS checklist
`
`‘Burning’ pain
`Hyperesthesia
`Temperature asymmetry
`Color changes
`Sweating changes
`
`Edema
`Nail changes
`Hair changes
`Skin changes
`Weakness
`Tremor
`Dystonia
`Decreased range of motion
`Hyperalgesia
`Allodynia
`
`Appendix C Proposed modified research diagnostic
`criteria for CRPS
`
`(1) Continuing pain which is disproportionate to any
`inciting event
`(2) Must report at least one symptom in each of the four
`following categories
`
`Sensory: reports of hyperesthesia
`Vasomotor: reports of temperature asymmetry and/or
`skin color changes and/or skin color asymmetry
`Sudomotor/edema: reports of edema and/or sweating
`changes and/or sweating asymmetry
`Motor/trophic: reports of decreased range of motion
`and/or motor dysfunction (weakness, tremor, dystonia)
`and/or trophic changes (hair, nail, skin)
`
`(3) Must display at least one sign in two or more of the
`following categories
`
`Sensory: evidence of hyperalgesia (to pinprick) and/or
`allodynia (to light touch)
`Vasomotor: evidence of temperature asymmetry and/or
`skin color changes and/or asymmetry
`Sudomotor/edema: evidence of edema and/or sweating
`changes and/or sweating asymmetry
`Motor/trophic: evidence of decreased range of motion
`and/or motor dysfunction (weakness, tremor, dystonia)
`and/or trophic changes (hair, nail, skin)
`
`References
`
`Amadio, P.C., Mackin