IASR
`
`PAINÒ 151 (2010) 870–876
`
`PAIN®
`
`w w w . e l s e v i e r . c o m / l o c a t e / p a i n
`
`Development of a severity score for CRPS
`R. Norman Harden a,⇑, Stephen Bruehl b, Roberto S.G.M. Perez c,d, Frank Birklein e, Johan Marinus d,f,
`
`Christian Maihofner g, Timothy Lubenow h, Asokumar Buvanendran h, Sean Mackey i, Joseph Graciosa a,
`Mila Mogilevski a, Christopher Ramsden a, Tanja Schlereth e, Melissa Chont b, Jean-Jacques Vatine j
`a Rehabilitation Institute of Chicago, Chicago, IL, USA
`b Vanderbilt University School of Medicine, Nashville, TN, USA
`c VU University Medical Center, Amsterdam, The Netherlands
`d Trauma Related Neuronal Dysfunction Consortium (TREND), The Netherlands
`e University Medical Center Mainz, Mainz, Germany
`f Leiden University Medical Center, Leiden, The Netherlands
`g University of Erlangen-Nuremberg, Erlangen, Germany
`h Rush University Medical Center, Chicago, IL, USA
`i Stanford University Medical Center, Stanford, CA, USA
`j Reuth Medical Center, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
`
`a r t i c l e
`
`i n f o
`
`a b s t r a c t
`
`Article history:
`Received 26 July 2010
`Received in revised form 17 September 2010
`Accepted 24 September 2010
`
`Keywords:
`Complex Regional Pain Syndrome
`Reflex Sympathetic Dystrophy
`CRPS
`RSD
`Psychometric
`SF-36
`Function
`Prediction
`CRPS Severity Score
`
`The clinical diagnosis of Complex Regional Pain Syndrome (CRPS) is a dichotomous (yes/no) categoriza-
`tion necessary for clinical decision-making. However, such dichotomous diagnostic categories do not
`convey an individual’s subtle and temporal gradations in severity of the condition, and have poor statis-
`tical power when used as an outcome measure in research. This study evaluated the validity and poten-
`tial utility of a continuous type score to index severity of CRPS. Psychometric and medical evaluations
`were conducted in 114 CRPS patients and 41 non-CRPS neuropathic pain patients. Based on the pres-
`ence/absence of 17 clinically-assessed signs and symptoms of CRPS, an overall CRPS Severity Score
`(CSS) was derived. The CSS discriminated well between CRPS and non-CRPS patients (p < .001), and dis-
`played strong associations with dichotomous CRPS diagnoses using both IASP diagnostic criteria
`(Eta = 0.69) and proposed revised criteria (Eta = 0.77–0.88). Higher CSS was associated with significantly
`higher clinical pain intensity, distress, and functional impairments, as well as greater bilateral tempera-
`ture asymmetry and thermal perception abnormalities (p’s < .05). In an archival prospective dataset,
`increases in anxiety and depression from pre-surgical baseline to 4 weeks post-knee arthroplasty were
`found to predict significantly higher CSS at 6- and 12-month follow-up (p’s < .05). Results indicate the
`CSS corresponds with and complements currently accepted dichotomous diagnostic criteria for CRPS,
`and support its validity as an index of CRPS severity. Its utility as an outcome measure in research studies
`is also suggested, with potential statistical advantages over dichotomous diagnostic criteria.
`Ó 2010 Published by Elsevier B.V. on behalf of International Association for the Study of Pain.
`
`1. Introduction
`
`The diagnosis of Complex Regional Pain Syndrome (CRPS; aka
`reflex sympathetic dystrophy, causalgia) has historically been a
`controversial issue, with internationally accepted diagnostic crite-
`ria only available since 1994 (published by the International Asso-
`ciation for the Study of Pain’s [IASP] committee on taxonomy) [32].
`Validation research over the past 10 years has led to a proposed
`
`⇑ Corresponding author. Address: Center for Pain Studies, Rehabilitation Institute
`
`of Chicago, 446 E. Ontario St., Suite 1011, Chicago, IL 60611, USA. Tel.: +1 312 238
`5654; fax: +1 312 238 7624.
`E-mail address: nharden@ric.org (R.N. Harden).
`
`revision of these diagnostic criteria (the ‘‘Budapest Criteria”
`[19,23]). Both these current and proposed CRPS diagnostic criteria
`result in dichotomous (yes/no) diagnostic decisions. While neces-
`sary for clinical decision-making, dichotomous diagnoses provide
`no information about individual differences in severity or lability
`of CRPS signs and symptoms and provide poor statistical power
`as an outcome measure. This latter issue may impact particularly
`on prospective studies necessary to identify factors contributing
`to CRPS development, given the relatively infrequent occurrence
`of CRPS following injury. Availability of a well-validated continu-
`ous CRPS score that corresponds with dichotomous CRPS diagnoses
`might facilitate CRPS research and provide a simple means of
`communicating the severity of CRPS both inter-patient and intra-
`patient over time.
`
`0304-3959/$36.00 Ó 2010 Published by Elsevier B.V. on behalf of International Association for the Study of Pain.
`doi:10.1016/j.pain.2010.09.031
`
`Grun. Exh. 1056
`PGR for U.S. Patent No. 9,408,862
`
`

`

`R.N. Harden et al. / PAINÒ 151 (2010) 870–876
`
`871
`
`Various proposals for CRPS scores have been described in the
`literature for more than 20 years [4,12,14,15,18,28,29,42,44,45].
`These were often proposed as a method of CRPS diagnosis, with
`‘‘cut scores” on these continuous measures suggested as a criterion
`for making clinical diagnostic decisions (e.g., [18,15,44]). A weak-
`ness common to most such scoring systems was that they were
`rationally-derived, and not followed up with systematic empirical
`validation to document their appropriateness or utility.
`the
`The best-validated continuous CRPS index available,
`Impairment level SumScore (ISS), was designed specifically to
`assess CRPS-related impairment [33,35]. The ISS has several compo-
`nents, including a visual analog scale and McGill Pain Questionnaire
`pain ratings, goniometric measures of active range of motion,
`temperature asymmetry as reflected in infrared (IR) thermometry,
`and edema assessed via volumetry. Assessed values for each compo-
`nent are transformed into 1–10 scores, which are then combined into
`a total ISS score. Although the ISS has proven to be a useful outcome
`measure in some studies (e.g., [41]), it was not intended to be a sum-
`mary measure of CRPS severity. As a result, it does not reflect several
`clinical features included in the current IASP diagnostic criteria for
`CRPS (e.g., hyperalgesia, allodynia, sweating and skin color changes),
`and reflects even less the more extensive features included in the
`proposed ‘‘Budapest Criteria” [19,21,23]. The need for quantitative
`testing and to transform these quantitatively-derived values into a
`scaling system before summing the components may make available
`scoring systems cumbersome for routine use [12,14,18,33].
`The current project sought to empirically test the validity of a
`continuous-type CRPS Severity Score (CSS) that better reflects fea-
`tures of CRPS included in current and proposed diagnostic criteria,
`and that can be calculated by any clinician based solely on a series
`of simple ‘‘bedside” tests. It was hoped that a severity score reflect-
`ing CRPS diagnostic features normally identified during standard
`clinical history and physical examinations might prove useful as
`an overall index of CRPS severity for both communicating clinical
`status and for research purposes.
`
`2. Methods
`
`2.1. Design
`
`A prospectively-obtained, international, multi-site sample was
`evaluated in a cross-sectional design to assess the potential utility
`of a continuous-type CRPS score as an index of CRPS severity.
`
`2.2. Subjects
`
`Subjects included a series of 114 CRPS patients who presented
`for evaluation and treatment at the data collection sites. All CRPS
`patients met published IASP diagnostic criteria for the disorder
`[32]. Both CRPS-I (80.7% of the sample) and CRPS-II (19.3%)
`patients were included in the study, with this diagnostic distinc-
`tion reflecting the clinical absence versus presence (respectively)
`of ‘‘major” (not defined) peripheral nerve injury. Both groups
`otherwise met identical IASP criteria for CRPS. Presence of nerve
`injury was confirmed by EMG/NCV testing in 86% of the CRPS-II
`patients in this study. Fracture was the single most common initi-
`ating event in the CRPS group (36.3%), with surgery and crush inju-
`ries contributing in an additional 33.7%. Distribution of CRPS
`patients across the study sites was: Reuth Medical Center (Israel;
`32.5%), University of Erlangen-Nuremberg (Germany; 12.3%), VU
`University Medical Center (Netherlands; 12.3%), Rehabilitation
`Institute of Chicago (US; 11.4%), Leiden University Medical Center
`(Netherlands; 11.4%), University Medical Center Mainz (Germany;
`8.8%), and Rush University Medical Center (US; 6.1%).
`A group of 41 non-CRPS neuropathic pain patients (i.e., neuro-
`pathic pain in the limbs without significant autonomic features)
`
`presenting for evaluation and treatment at the study sites were
`also included in the study for two purposes. First, the non-CRPS
`comparison group permitted examination of the continuous-type
`CRPS Severity Score in terms of its correspondence with dichoto-
`mous CRPS/non-CRPS diagnoses using the IASP criteria and pro-
`posed modified ‘‘Budapest Criteria” [19]. Second, given the
`absence of low CRPS Severity Scores in the CRPS patient group
`(see below; the lowest CRPS Severity Score was 7 on a 0–17 scale
`in the CRPS group), availability of the non-CRPS group insured that
`a less restricted range of CRPS Severity Scores was available for
`evaluating the utility of these scores as an index of CRPS severity
`in correlational analyses. That is, to demonstrate that higher CRPS
`scores are associated with greater pain, distress, and functional
`impairments, it was also necessary to show that low CRPS Severity
`Scores were associated with lower levels on these measures. The
`non-CRPS patient group underwent evaluation procedures identi-
`cal to the CRPS group. Diagnoses in the non-CRPS group included
`peripheral neuropathy in a single extremity isolated to a specific
`nerve distribution (45%), radiculopathy (30%), diabetic peripheral
`neuropathy (15%), and carpal or tarsal tunnel syndrome (10%).
`Non-CRPS neuropathic pain disorders were diagnosed by the pres-
`ence of spontaneous pain with clear neuropathic etiology sup-
`ported by relevant testing where appropriate (e.g., EMG and
`clinical examination consistent with pain and symptoms restricted
`to a specific peripheral nerve distribution following known injury
`to that nerve, extremity pain coexisting with known diabetes mel-
`litus, pain in a radicular pattern with disk herniation confirmed by
`MRI, etc.). Distribution of non-CRPS patients across the study sites
`was: University Medical Center Mainz (Germany; 24.4%), Leiden
`University Medical Center (Netherlands; 19.5%), VU University
`Medical Center (Netherlands; 19.5%), Reuth Medical Center (Israel;
`14.6%), Rush University Medical Center (US; 12.2%), and Stanford
`University Medical Center (US; 9.8%). Summary data regarding
`the two groups of study patients are presented in Table 1. Addi-
`tional data regarding the samples are provided in related work
`[21].
`
`2.3. Procedures
`
`Subjects provided written informed consent, and then com-
`pleted the Rand-36 Health Survey to characterize their pain, emo-
`tional state, and functional level [43]. Next, for all patients in both
`groups, a study physician conducted an evaluation of CRPS signs
`and symptoms using the CRPS checklist detailed below (results
`for both patient groups are summarized in Table 2). This involved
`obtaining a patient history to assess symptoms, as well as conduct-
`ing a standardized physical examination to assess signs. Both a
`standard protocol and a video example of examination procedures
`were provided to all sites to enhance consistency. To better charac-
`terize range of motion limitations identified during the physical
`examination, active range of motion was objectively assessed using
`a standard goniometer provided to all study sites. Flexion and
`extension in the affected elbow/knee (upper/lower extremity pain)
`and affected wrist/ankle (upper/lower extremity pain) were evalu-
`ated by this quantitative range of motion examination. To better
`characterize the degree of temperature asymmetry noted on the
`physical examination, temperatures in the center of the affected
`hand (palmar surface) or foot (plantar surface) and the contralat-
`eral hand/foot were determined while in a room temperature envi-
`ronment (minimum 30 min of acclimatization) using standard
`infrared (IR) thermometers provided to all study sites (Exergen
`Corp., Watertown, MA). This simple temperature assessment
`methodology was used in an attempt to provide quantitative con-
`firmation of clinically-discerned temperature asymmetry.
`Thermal Quantitative Sensory Testing (tQST; Medoc TSA-II,
`Medoc Inc, Tel Aviv, Israel) was also carried out for patients at
`
`

`

`872
`
`R.N. Harden et al. / PAINÒ 151 (2010) 870–876
`
`Table 1
`Sample characteristics by diagnostic subgroup.
`
`Variable
`
`Diagnosis
`
`CRPS
`(n = 114)
`
`63.0
`40.5 ± 15.74
`19.1 (52.74)
`48.7
`55.8
`28.3 ± 3.28
`
`37.2 ± 3.48
`
`Gender (female%)*
`Age (years)**
`Pain duration (median (IQR) in months)*
`Affected extremity (% lower extremity)**
`Affected side (% right)
`Affected side cold perception threshold
`(°C)*
`Affected side warmth perception threshold
`(°C)**
`Affected side heat pain threshold (°C)**
`42.8 ± 4.19
`Mean asymmetry by IR thermometry (°C)* 0.62 ± 1.82
`111.9 ± 37.37
`Affected side AROM – elbow/knee flexion
`(°)*
`Affected side AROM – elbow/knee
`extension (°)
`Affected side arom – wrist/ankle flexion (°)
`Affected side AROM – wrist/ankle
`extension (°)
`Rand 36 – physical functioning
`Rand 36 – role limitations – physical
`health
`Rand 36 – role limitations – emotional
`problems
`Rand 36 – energy/fatigue
`Rand 36 – emotional well-being
`Rand 36 – social functioning
`Rand 36 – Pain
`Rand 36 – general health
`CRPS Severity Score**
`
`10.8 ± 45.41
`
`37.2 ± 33.91
`35.2 ± 30.39
`
`41.4 ± 25.03
`15.0 ± 30.20
`
`43.7 ± 45.75
`
`37.3 ± 22.41
`56.2 ± 21.35
`48.0 ± 26.73
`28.1 ± 20.10
`50.80 ± 20.83
`13.3 ± 2.35
`
`Non-CRPS
`(n = 41)
`
`41.5
`52.6 ± 15.15
`40.9 (95.78)
`74.3
`58.3
`25.5 ± 4.98
`
`41.1 ± 3.71
`
`46.0 ± 2.41
`0.13 ± 0.98
`127.3 ± 28.57
`
`8.4 ± 36.54
`
`39.0 ± 32.39
`41.6 ± 34.58
`
`41.8 ± 25.29
`23.8 ± 37.53
`
`46.2 ± 44.34
`
`39.5 ± 22.70
`60.3 ± 24.76
`52.8 ± 28.66
`33.2 ± 21.24
`49.5 ± 23.64
`7.0 ± 3.34
`
`total of 58 CRPS patients and 13 Non-CRPS patients. A standardized
`protocol was used across all study sites obtaining these data. The
`tQST protocol employed a computer-controlled 30  30 mm Peltier
`thermistor probe that was used to evaluate cold and warmth per-
`ception thresholds and heat pain threshold (mean of 3 trials each)
`using the ‘method of limits’ program standard on the device. For
`upper extremity CRPS, the probe was placed sequentially on three
`adjacent sites on the volar forearm of the affected extremity. For
`lower extremity CRPS, the probe was similarly placed on three
`adjacent sites on the dorsal mid-calf. Prior to each trial, the probe
`was maintained at an adaptation temperature of 32 °C.
`All study procedures were approved by the relevant ethical re-
`view boards at participating institutions.
`
`2.4. Measures
`
`2.4.1. CRPS database checklist
`In order to insure standardized assessment of signs and symp-
`toms across study sites, a CRPS database checklist similar to that
`used in our past multi-site research work was employed [7,8,20].
`This checklist presented a complete list of the signs and symptoms
`traditionally used to diagnose CRPS, as well as other signs/symptoms
`(e.g., trophic changes, motor abnormalities) reported to be associ-
`ated with the disorder in previous literature but not incorporated
`in the IASP diagnostic criteria [27,32,36,38,39,42,40]. These latter
`signs and symptoms are included in the proposed Budapest diagnos-
`tic criteria [21]. Categorical measures (e.g., presence or absence)
`were used to assess all signs and symptoms because of the potential
`for decreased inter-rater reliability using interval rating scales
`[26,34]. Written standardized procedures and an instructional video
`to demonstrate the data collection procedures were provided with
`the checklist to maximize uniform assessment across sites. Investi-
`gators at all sites were highly proficient in English, thereby minimiz-
`ing the potential impact of language issues. Copies of the database
`checklist and instructions are available from the authors.
`
`2.4.2. Rand-36 Health Survey
`The Rand-36 Health Survey consists of the same 36 items in-
`cluded in the SF-36 questionnaire frequently employed in medical
`outcomes research [43]. The Rand-36, however, employs a simpler
`unweighted scoring system for these items [Rand Health Sciences
`Program, 1992]. Validated versions of the questionnaire were
`available in German [10], Hebrew [30], and Dutch [1] for use with
`patients at non-English speaking data collection sites. The Rand-36
`assesses eight health-related areas including pain, general health,
`and the impact of pain or disease in six specific domains of life
`function. Each scale falls in 0–100 range, with lower scores indicat-
`ing greater pain or dysfunction.
`
`2.4.3. CRPS Severity Score
`To provide a continuous type quantitative index of the signs and
`symptoms of CRPS, the CRPS Severity Score (CSS) was created. For
`each patient, the history and physical examination as recorded on
`the CRPS Database Checklist was coded so that 1 = presence and
`0 = absence for each of 17 diagnostic CRPS features (see Table 2).
`Scores for each of these CRPS features were then summed (un-
`weighted) to create the overall CRPS Severity Score. Signs and
`symptoms were included as separate elements in this score given
`work showing that they do not necessarily correspond, due in
`some part to the subjectivity of historical symptoms and to the
`lability of CRPS signs which may result in relevant clinical features
`being missed on any ‘spot’ physical examination [20]. Self-reported
`CRPS symptoms included in the CSS were hyperpathia/allodynia
`(e.g., increased or prolonged pain to a mildly noxious stimuli,
`mechanical or thermal allodynia [pain to normally innocuous stim-
`uli]), bilateral temperature asymmetry, skin color changes, edema,
`
`Note: Summary statistics are presented as percentages or mean ± SD. A negative
`value for mean asymmetry by IR thermometry indicates the affected side was
`cooler. Lower Rand 36 scores indicate greater pain, distress, or dysfunction. IQR,
`interquartile range; AROM, active range of motion.
`* p < .05.
`** p < .01.
`
`the study sites in Israel and Germany, as well as the Rehabilitation
`Institute of Chicago and Stanford sites (data were only obtained for
`a subset of subjects at these latter two sites due to equipment
`availability and time constraints). tQST data were available for a
`
`Table 2
`Diagnostic signs and symptoms included in the CRPS Severity Score by subgroup.
`
`Variable
`
`Self-reported symptoms (% yes)
`Allodynia, Hyperpathia**
`Temperature asymmetry**
`Skin color asymmetry**
`Sweating asymmetry**
`Asymmetric edema**
`Trophic changes**
`Motor changes**
`Decreased active range of motion**
`
`Signs observed on examination (% yes)
`Hyperpathia to pinprick**
`Allodynia**
`Temperature asymmetry by palpation**
`Skin color asymmetry**
`Sweating asymmetry**
`Asymmetric edema**
`Trophic changes**
`Motor changes**
`Decreased active range of motion**
`
`* p < .05.
`** p <.01.
`
`Diagnosis
`
`CRPS
`
`Non-CRPS
`
`88.6
`87.7
`90.4
`63.2
`88.6
`75.4
`89.5
`86.8
`
`78.9
`71.1
`64.9
`84.2
`38.6
`64.9
`69.3
`78.1
`86.0
`
`65.9
`36.6
`29.3
`14.6
`41.5
`39.0
`46.3
`34.1
`
`43.9
`29.3
`14.6
`34.1
`12.2
`26.8
`31.7
`39.0
`36.6
`
`

`

`R.N. Harden et al. / PAINÒ 151 (2010) 870–876
`
`873
`
`sweating asymmetry, trophic/dystrophic changes (hair, nails, or
`skin), motor changes (e.g. weakness,
`tremor, dystonia), and
`decreased active range of motion. CRPS signs (present on examina-
`tion at the time of data collection) that were included in the CSS
`were hyperpathia/hyperalgesia to pinprick, allodynia (to light
`touch [brush], cold, warm, vibration, or deep manual joint pres-
`sure), temperature asymmetry, skin color changes, edema, sweat-
`ing asymmetry, trophic/dystrophic changes, motor changes, and
`decreased active range of motion. The resulting CSS ranged poten-
`tially from 0 to 17, with higher scores indicating greater CRPS
`severity. As an example, a patient self-reporting hypersensitivity
`to touch, edema, skin color changes, and weakness who displays
`only hyperalgesia and edema ‘objectively’ on examination would
`have a CRPS Severity Score of 6. In the combined patient groups
`the scores ranged from 2 to 17 (overall mean ± SD = 11.7 ± 3.84;
`range 7–17 if CRPS subjects only). Internal consistency reliability
`of the 17 items comprising the CRPS Severity Score was high
`(alpha = 0.88), indicating that combining all items into a single
`summary score was justified.
`
`2.5. Statistical analysis
`
`Analyses were conducted using PASW Statistics 18 for Windows
`(SPSS Inc., Chicago, IL). A number of approaches were used to val-
`idate and evaluate the potential utility of the CSS derived in this
`study. First, its association with known CRPS and non-CRPS diag-
`nostic groups was examined. These analyses examined CRPS as
`diagnosed according to the current IASP diagnostic criteria [32]
`and according to the proposed Budapest clinical and research crite-
`ria for CRPS [19,21]. Differences in the CSS between CRPS and non-
`CRPS diagnostic groups were examined using t-tests as an index of
`discriminative validity. Strength of associations between continu-
`ous CSS and dichotomous diagnostic categories was also examined
`using the nonlinear correlation coefficient, Eta. A second analytic
`approach was to evaluate the associations (using Pearson correla-
`tion coefficients) between CSS and psychometric measures reflect-
`ing pain, emotional status, and function, as well as objective
`measures of temperature asymmetry, active range of motion, and
`sensory function (tQST). These analyses were designed to help
`demonstrate the concurrent validity of the CSS. It was hypothe-
`sized that higher CSS values, if a valid index of CRPS severity,
`would be associated with greater pain, distress, and dysfunction,
`more impaired range of motion, and more abnormal sensory func-
`tion. Finally, the CSS as described above was derived in an archival
`prospective total knee arthroplasty dataset detailed fully in our
`previous work (see [21,22]), and its utility as a CRPS outcome mea-
`sure was evaluated. Specifically, given our previous work showing
`in a prospective design that pre-surgical anxiety was a significant
`predictor of early post-surgical dichotomous CRPS diagnoses [22],
`we re-evaluated the ability of pre-surgical anxiety and depression
`and early post-surgical changes in these psychological factors to
`predict longer-term CRPS sign and symptom burden as reflected
`in continuous CSS values at 6 months (n = 76) and 12 months
`(n = 65) post-surgery. These analyses were designed in part to
`demonstrate the potential advantages of continuous over dichoto-
`mous measures of CRPS in the research context due to enhanced
`statistical power. These latter analyses used linear regression
`procedures.
`
`3. Results
`
`according to IASP criteria was nearly double that in patients with
`non-CRPS neuropathic pain diagnoses, a difference that was statis-
`tically significant [t(153) = 13.10, p < .001]. The continuous type
`CSS demonstrated large associations with dichotomous CRPS diag-
`nostic categorizations. Using an IASP diagnosis of CRPS as the
`dependent measure, the magnitude of association with continuous
`CSS values was Eta = 0.69 (similar to other correlation coefficients,
`Eta ranges potentially from 0 to 1.00 to indicate increasing magni-
`tude of association). Using the Budapest research diagnostic crite-
`ria as the dependent measure, the association with continuous CSS
`values was larger (Eta = 0.77). Finally, associations with the Buda-
`pest clinical criteria showed the strongest association (Eta = 0.88),
`indicating 77% shared variance between the dichotomous and
`continuous-type CRPS measures.
`
`3.2. Concurrent associations between CRPS Severity Scores and
`outcome indices
`
`Table 3 summarizes the pattern of zero-order correlations
`between CSS values and outcome indices obtained at the time of
`the history and physical examination from which data for deriving
`CSS values were obtained. As predicted, higher CSS values were
`associated on the Rand-36 measure with significantly higher pain
`intensity, worse physical and social functioning, greater role limi-
`tations due to physical and emotional problems, lower energy le-
`vel, and lower emotional well-being (i.e., greater distress). These
`findings corroborated the hypothesis that the CSS is a valid index
`of CRPS severity, in terms of deleterious effects on psychosocial
`function across several key domains. CRPS Severity Scores were
`not associated significantly with the Rand-36 measure of general
`health, indicating that the CSS is not simply a surrogate measure
`of overall physical health.
`Higher CSS values were associated with significantly greater
`quantitative temperature asymmetry (affected extremity colder)
`based on the IR thermometry measure. Results of thermal QST
`evaluation indicated that higher CSS values were also associated
`with significantly higher cold perception threshold (p < .05) and
`lower warmth perception threshold (p < .01), as well as signifi-
`cantly greater heat hyperalgesia (p < .01). Finally, higher CSS values
`were associated with significantly lower elbow/knee flexion
`
`Table 3
`Zero-order correlations between CRPS Severity Scores and concurrent outcome
`indices.
`
`Outcome index
`
`Rand 36 – physical functioning
`Rand 36 – role limitations – physical health
`Rand 36 – role limitations – emotional problems
`Rand 36 – energy/fatigue
`Rand 36 – emotional well-being
`Rand 36 – social functioning
`Rand 36 – Pain
`Rand 36 – general health
`Affected side cold perception threshold (°C)
`Affected side warmth perception threshold (°C)
`Affected side heat pain threshold (°C)
`Temperature asymmetry by IR thermometry (°C)
`Affected side AROM – elbow/knee flexion (°)
`Affected side AROM – elbow/knee extension (°)
`Affected side AROM – wrist/ankle flexion (°)
`Affected side AROM – wrist/ankle extension (°)
`
`CRPS Severity
`Score correlation
`0.19*
`0.25**
`0.22*
`0.20*
`0.19*
`0.27**
`0.38**
`0.03
`0.26*
`0.31**
`0.23 
`0.22*
`0.23**
`0.02
`0.11
`0.20*
`
`3.1. Associations between CRPS Severity Scores and dichotomous CRPS
`diagnoses
`
`Table 1 compares CSS values between CRPS and non-CRPS
`diagnostic groups. The mean CSS in patients diagnosed with CRPS
`
`Note: Negative correlations for Rand 36 scores indicate that higher CRPS scores were
`associated with greater pain, distress, and dysfunction. The negative correlation for
`temperature asymmetry indicates that higher CRPS scores were associated with
`greater asymmetry in which the affected side was colder.
`  p < .10.
`* p < .05.
`** p < .01.
`
`

`

`874
`
`R.N. Harden et al. / PAINÒ 151 (2010) 870–876
`
`(p < .01) and wrist/ankle extension (p < .05) on goniometric testing,
`consistent with functional deficits noted on the Rand-36.
`
`3.3. CRPS Severity Scores as a prospective CRPS outcome measure
`
`In a prospective study, we found that pre-surgical anxiety but
`not depression significantly predicted dichotomous CRPS diagno-
`ses at 4 weeks post-surgery in total knee arthroplasty patients
`[22]. Due in part to the low statistical power related to use of a
`dichotomous diagnostic measure and subject attrition over time
`in this previous study, there were no significant findings for psy-
`chological predictors at more extended follow-up [22]. This dataset
`was re-analyzed using continuous CSS values as described in the
`current study as an alternative CRPS outcome measure. Although
`baseline (pre-surgical) anxiety (State Trait Anxiety Inventory
`[37]) and depression (Beck Depression Inventory [3]) did not pre-
`dict CRPS sign and symptom burden as reflected in CSS values at
`6- or 12-month follow-up, early post-surgical changes in these
`measures (from pre-surgical baseline to 4 weeks post-surgery)
`were found to significantly predict CSS values at extended follow-
`up. Specifically, greater early post-surgical increases in depressive
`symptoms predicted higher CSS values at both 6-month follow-up
`[beta = 0.24;
`t(73) = 2.10, p < .05]
`and 12-month follow-up
`[beta = 0.26; t(62) = 2.06, p < .05]. Similarly, greater early post-sur-
`gical
`increases in anxiety predicted higher CSS values at
`6-month follow-up [beta = 0.25; t(73) = 2.16, p < .05], although it
`failed to predict CSS status at 12-month follow-up (p > .10). It was
`notable that CSS values at 6-month follow-up were also positively
`associated with concurrent clinical pain intensity as reflected in
`McGill Pain Questionnaire-Short Form total scores (r = 0.40,
`p < .001) [31]. Similar associations at the 12-month follow-up were
`in the same direction but failed to achieve statistical significance
`(r = 0.22, p < .11). For direct comparison of continuous type versus
`dichotomous CRPS measures, logistic regression analyses modeled
`after those above were conducted using dichotomous IASP CRPS
`diagnoses as the outcome measure. These analyses did not reveal
`significant predictive effects for baseline or early post-surgical
`changes in anxiety or depression at long-term follow-up (p’s > .10).
`
`4. Discussion
`
`Dichotomous (yes/no) CRPS diagnoses are necessary for clinical
`decision-making, but do not adequately capture differences in the
`severity or lability of the CRPS, the clinical presentation of which
`can be quite variable between patients or over time within a given
`patient [7,13]. Dichotomous diagnostic schemes also have inher-
`ently poor statistical power when used as outcome measures in re-
`search studies, a weakness particularly important in prospective
`studies of CRPS following injury, given the relative infrequency of
`the syndrome. A variety of continuous-type CRPS scores have been
`proposed, although few have been subjected to adequate empirical
`validation [4,12,14,15,18,28,29,42,44,45]. One continuous CRPS-re-
`lated measure has been well-validated (ISS; [33,35]). However,
`because the ISS was intended to index CRPS-related impairment,
`it does not capture the full spectrum of CRPS-related signs and
`symptoms incorporated in accepted or proposed CRPS diagnostic
`criteria [23,32]. The current study sought to pilot the development
`and validation of a simple continuous CRPS Severity Score that
`does not require specialized testing, training or equipment, and
`which corresponds well with the current and proposed dichoto-
`mous diagnostic criteria. This continuous-type CRPS Severity Score
`(CSS) may be useful to communicate CRPS severity efficiently be-
`tween clinicians and monitor intra-patient status over time, and
`have substantial statistical advantages over dichotomous CRPS
`diagnoses as an outcome measure in CRPS research studies.
`
`The 17 signs and symptoms comprising this CSS (directly de-
`rived from the proposed ‘‘Budapest Criteria” [19,21,23]) were
`found to have high internal consistency reliability, supporting their
`use as components of a single summary score. The validity of the
`CSS was demonstrated in several ways. As designed, it discrimi-
`nated well between CRPS and non-CRPS diagnostic groups.
`Moreover, it demonstrated a strong association with diagnoses
`determined based on current IASP diagnostic criteria for CRPS
`[32], and an even stronger association with the revised Budapest
`diagnostic criteria that have been proposed, especially the ‘‘clini-
`cal” criteria [19,21,23]. The CSS demonstrated a pattern of signifi-
`cant positive associations with pain intensity, emotional distress,
`and impaired function as would be expected if it were a sensitive
`index of CRPS severity. In addition, higher scores were associated
`with reduced active range of motion and greater bilateral temper-
`ature asymmetry as was predicted. The results above support the
`discriminative and concurrent validity of the CSS.
`An existing prospective archival dataset was also used to test
`the sensitivity of the CSS as an outcome measure reflecting CRPS.
`We had previously shown that pre-surgical anxiety (but not
`depression) was a significant predictor of dichotomous CRPS diag-
`noses at 4 weeks following total knee arthroplasty, but not at long-
`er-term follow-up [22]. Results of our re-analysis of this archival
`dataset revealed that greater increases in anxiety and depression
`from pre-surgical baseline to 4 weeks post-surgery were a signifi-
`cant predictor of CSS values at 6-month (anxiety and depression)
`and 12-month follow-up (depression). In contrast, neither affective
`measures significantly predicted dichotomous IASP CRPS diagnoses
`at either follow-up. This latter finding highlights the statistical
`advantages of a continuous-type CRPS index over traditional
`dichotomous diagnostic outcomes. Given the known positive asso-
`ciations between psychological distress, catecholamine release,
`and s