Classification of Chronic Pain Associated With Spinal Cord Injuries

Transcription

1 1708 Classification of Chronic Pain Associated With Spinal Cord Injuries Diana D. Cardenas, MD, MHA, Judith A. Turner, PhD, Catherine A. Warms, MSN, Helen M. Marshall, BS ABSTRACT. Cardenas DD, Turner JA, Warms CA, CHRONIC PAIN IN PATIENTS with spinal cord injury Marshall HM. Classification of chronic pain associated with (SCI) remains a common clinical problem that is difficult spinal cord injuries. Arch Phys Med Rehabil 2002;83: to manage. The several published definitions and categorizations of pain are confusing and make clinical applications difficult. Investigators have proposed different categorization schemes based on the presumed site of pathology, location of pain, temporal course, or clinical presentation. One of the simplest and earliest schemes divides pain into 3 categories: pain above the spinal cord lesion, pain at the level of the lesion, Objectives: To determine interrater reliability of a classification system for chronic pain in persons with spinal cord injury (SCI) and to determine the frequency and characteristics of various pain types as categorized by this system. Design: Independent categorization (based on questionnaires; for 15 persons, questionnaires plus personal interviews) by 2 investigators. Setting: Community. Participants: A total of 163 individuals aged 18 years with SCI and pain. Interventions: Not applicable. Main Outcome Measures: Pain categories, Short-Form McGill Pain Questionnaire, and Chronic Pain Grade questionnaire. Results: Among 41 (25%) questionnaires categorized independently by 2 investigators, strength of agreement in categorizing 68 pain problems was substantial (.68). For 15 persons whose pain was categorized in person by 2 investigators, strength of agreement was also substantial (.66). Among 163 survey respondents with pain, the most common worst pain was SCI pain (31.9%). Mean characteristic pain intensity standard deviation for worst pain, regardless of type, was on a scale from 0 to 100. On average, for worst pain, respondents reported moderate pain-related disability ( ; scale range, 0 100). Although certain pain descriptors were more often associated with a specific type of pain, none was pathognomonic. Conclusions: Substantial interrater reliability was achieved in determining pain categories by use of responses to a questionnaire with a classification system based on presumed pathology. Adding interviews with patients increased our ability to classify pain but did not improve overall interrater reliability. Key Words: Pain; Rehabilitation; Spinal cord injuries by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the Departments of Rehabilitation Medicine (Cardenas, Turner, Warms, Marshall) and Psychiatry and Behavioral Sciences (Turner), University of Washington, Seattle, WA. Supported by the National Institute of Child Health and Human Development, National Institute of Neurological Disorders and Stroke, National Institutes of Health (grant no. 1 PO1 HD/NS33988), and the Office of Special Education and Rehabilitation Service, National Institute on Disability and Rehabilitation Research, US Department of Education (grant no. H133N000003); conducted through the Clinical Research Center facility at the University of Washington, supported by the National Institutes of Health (grant no. M01-RR-0037). No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the author(s) or upon any organization with which the author(s) is/are associated. Reprint requests to Diana D. Cardenas, MD, MHA, University of Washington Sch of Medicine, Dept of Rehabilitation Medicine, Box , Seattle, WA 98195, dianamac@u.washington.edu /02/ $35.00/0 doi: /apmr and pain below the lesion. 1 Although this system has the advantage of being simple, it does not address pathology. Difficulties also arise with such a classification system when the patient has an incomplete lesion, creating the potential for both musculoskeletal pain below the lesion, such as in a partially functional limb, and pain secondary to the lesion itself. Some classification systems use broad terms to describe pathology, such as musculoskeletal, visceral, radicular, orsci pain. Guttmann 2 wrote about pain above the lesion resulting from soft tissue pathology; pain at or above the lesion because of root or segmental irritation; phantom sensations; and pain arising from viscera. Guttmann used the term phantom sensations to describe pain below the level of the lesion, but the term phantom pain should be used only in association with the loss of an extremity. Burke and Woodward 3 described 3 categories of pain: root pain, visceral pain, and pain occurring distal to the lesion. This classification system did not include musculoskeletal pain or pain at the level of the lesion. Donovan et al 4 used the following terms: segmental nerve pain, spinal cord pain, visceral pain, muscle tension or mechanical pain, and so-called supratentorial or psychogenic pain. More recently, Loubser and Donovan 5 proposed 2 broad categories: neurologic and nonneurologic pain. The neurologic category was subdivided into segmental, central, and visceral pain, and the nonneurologic category was subdivided into musculoskeletal and psychogenic problems. The overlapping nature of psychogenic problems with pain from any cause makes the category of psychogenic pain less attractive. Terminology remains confusing, particularly regarding the neuropathic pain associated with the spinal cord lesion itself that occurs below the level of the lesion, the most common type of chronic pain after SCI. 6-9 This pain has been called central pain, diffuse pain, SCI pain, dysesthetic pain syndrome, or phantom body pain. Some argue that the term central pain is not specific, because it can relate to both brain and spinal cord pathology. The term dysesthetic pain syndrome ties in the word dysesthesia, which describes an unpleasant abnormal sensation, whether spontaneous or evoked, 10 but such a sensation may be produced by peripheral as well as central nervous system lesions. Use of the term SCI pain to describe the pain of central origin avoids the problem of referring to the brain as well as the spinal cord. Siddall et al 11 proposed 4 categories of pain: musculoskeletal, visceral, neuropathic, and other, which he termed AXIS 1. Neuropathic pain was further subdivided into AXIS 2: neuropathic at-level or below-level pain. At the neuropathic level, pain was further subdivided into AXIS 3: radicular or central

2 PAIN CLASSIFICATION, Cardenas 1709 pain. 11 The other category included specific pains that are a consequence of SCI, such as headaches that may occur with autonomic dysreflexia. A scheme somewhat similar to Siddall s was suggested by Bryce and Ragnarsson. 12 In a more recent publication, Siddall et al 13 modified their system by using 2 broad categories nociceptive and neuropathic and subdividing nociceptive pain into musculoskeletal and visceral pain. Neuropathic pain was further subdivided into above-, at-, or below-level pain. No longer was there a need for an other category, because such pain as dysreflexic headache was placed under visceral, and specific neuropathic pains were grouped according to location in relation to the lesion. Of note, the term spinal cord injury pain was used by Siddall 13 to describe the pain in anesthetic regions below the level of injury. This brief review highlights the discrepancies in proposed classification schemes. The difficulty in understanding the mechanisms in the development of pain and the use of different clinical terminology based on symptoms add to the problem. Many investigators 13,14 have pointed out the problems associated with the lack of a universally accepted and applied classification system for pain in SCI. The problems include wide variation in estimates of pain prevalence 5,6,15,16 and the paucity of consistent reports in the literature about factors associated with chronic pain in SCI. For example, several studies reported lower rates of pain in persons with tetraplegia, 9, study 20 found pain to be more frequent among persons with cervical and lumbar injuries than persons with thoracic injuries, and 2 studies 21,22 found no relation between pain and neurologic level. The trend in the literature is for investigators to propose new classification systems in efforts to standardize the terminology. The classification system we propose has 2 major categories: neurologic and musculoskeletal. Neurologic pain is divided into 4 subcategories: SCI pain, transition zone pain, radicular pain, and visceral pain. Musculoskeletal pain is divided into mechanical spine pain (pain in the back or neck affected by activity and position) and overuse pain (often above the injury level in areas of normal sensation or sometimes below the injury level in incomplete injuries). SCI pain is that which is felt below the level of the lesion in an area without normal sensation. It may be related to activity and may be associated with allodynia. Transition zone pain occurs at the level of the lesion, is usually bilateral, and is not related to activity or affected by position, but it is associated with allodynia. Radicular pain may occur at any dermatomal level, is usually unilateral, usually radiates, and is related to activity and position. Visceral pain is felt in the abdomen and is not related to activity, affected by position, or associated with allodynia. Chronic visceral pain is rare, and no source of visceral pathology may be found. Although this system is similar in some ways to that proposed by Anke et al, 21 it differs in avoiding the use of pain qualifiers for categorization and includes additional categories for musculoskeletal pain. Anke 21 divided neurogenic pain into 4 categories: diffuse pain (for pain below the lesion), segmented pain, root pain, and visceral pain. To date, no universally accepted classification system has been adopted despite such efforts, and no information is available in the literature about the reliability of any of these schemes. This study evaluated a classification system based on presumed pathology and use of a standard set of questions and sought to determine the frequency and characteristics of various pain types as categorized by this system. METHODS Participants Study participants were solicited primarily through the mailing list of the newsletter of the Northwest Regional Spinal Cord Injury System (NWRSCIS), a comprehensive, interdisciplinary service delivery model system funded in part by the National Institute on Disability and Rehabilitation Research. Notices about the study were also placed in clinics in the Seattle, WA, area that serve patients with SCI and in the NWRSCIS newsletter. Questionnaires were mailed to 399 adults with SCI who were selected randomly from the mailing list (n 334) or who called in response to a notice (n 65). Each questionnaire was accompanied by a consent form and a cover letter inviting those aged 18 years with SCI to participate in the study. Each respondent was paid $20 for completing the survey. Both the survey and protocol were approved by the University of Washington Human Subjects Review Committee. Sixty percent of the questionnaires, excluding those mailed to addressees who were deceased, ineligible, or not at that address, were returned. Thus, 215 adults with SCI completed and returned the survey. A total of 167 surveys were returned by people on the mailing list (response rate, 50%), and 48 were returned by people who called in (response rate, 74%). Of the total group, 163 (75.8%) persons reported that they were currently experiencing pain. Sociodemographic and other characteristics of the 163 persons with pain are listed in table 1. To assess the extent to which this sample was representative of the Pacific Northwest population of persons with SCI, we compared it with 961 persons registered in the NWRSCIS database. In our sample, there was a lower proportion of men (69.9% vs 79%); a comparable proportion of white (84.7% vs 83%), Hispanic/ Chicano (4.3% vs 3.2%), and Asian, Pacific Islander, and mixed race (2.3% vs 2.8%) participants; a slightly higher proportion of Native Americans (3.7% vs 2.1%); and a slightly lower proportion of black participants (3.6% vs 4.9%). Eightyfive persons (52%) had tetraplegia, 76 (47%) had paraplegia, and the level of injury was unknown in 2 persons. Subjects were highly educated. Only 8% had less than a high school education, and 32.5% had some college education. Of the total sample, 33.1% had either a college degree or graduate or professional school education, 25.8% were employed (17.2% full-time, 8.6% part-time), 8% were in school full-time, 14.1% were retired, and 6.1% were homemakers. Exactly 50.9% said that they were unemployed because of their disability, and 13.5% said that they were unemployed because of pain. Measures The questionnaire, which was pilot tested, consists of 34 questions, 7 of which contain multiple questions. Thus, there were 188 items in the questionnaire. The questionnaire assessed sociodemographic variables (age, sex, education, marital status) and information about the SCI (date of injury, etiology, level of injury, completeness). The questionnaire also asked whether respondents had a current pain problem, and, if the answer was yes, they were questioned about their pain. Subjects with more than 1 pain problem were asked to identify their worst pain, called Pain 1, and second worst pain, called Pain 2, and to answer questions about each of the pains, including their duration. On a body diagram, they were asked to shade in the location of each pain as above, at, or below the level of injury. Subjects identified the presumed source of each pain (musculoskeletal vs nervous system) and indicated whether pain worsened with activity, position, change of position, or light touch. This information was used to determine

3 1710 PAIN CLASSIFICATION, Cardenas Table 1: Sample Characteristics (N 163) Gender Men 69.9 Women 30.0 Mean age SD (range) (y) Men (18 77) Women (18 76) Mean years since SCI SD (5 33) Ethnic group White 84.7 Hispanic/Chicano 4.3 Black 3.6 Native American 3.7 Asian 1.2 Pacific Islander/mixed race 1.2 Marital status Married 33.7 Living with partner 10.4 Never married 34.4 Separated/divorced 18.5 Education (highest level) Grade 11 or lower 8.0 High school/ged 20.2 Vocational/technical/business school 6.1 Some college 32.5 College graduate 18.4 Graduate/professional school 14.7 Employment Employed full-time 17.2 Employed part-time 8.6 School/vocational training 8.6 Retired 14.1 Homemaker 6.1 Unemployed due to disability 50.9 Unemployed due to pain 13.5 Cause of SCI Motor vehicle crash 49.1 Fall 19.6 Abbreviation: SD, standard deviation; GED, General Education Development. pain categorization. Other information used for pain categorization was level and completeness of injury, according to the subject. (The questions we developed regarding pain categorization are available on request.) The survey also included the 7-item Chronic Pain Grade (CPG) questionnaire, which assesses pain intensity and its interference with normal daily activities. 23 Subjects were asked to complete the items for each pain problem. The CPG questionnaire has validity and high internal consistency and has been used in previous mail survey instruments Characteristic pain intensity was calculated by averaging 0 to 10 ratings of current pain, worst pain in the past 3 months, and average pain in the past 3 months, and then multiplying by Pain-related disability scores were calculated by averaging 0 to 10 ratings of pain interference with daily, social, and work and housework activities in the past 3 months and then multiplying by The CPG classifies individuals with pain into 4 categories: I is low pain intensity and low pain-related disability, II is high pain intensity and low pain-related disability, III is moderate pain-related disability, and IV is severe pain-related disability. One item was added to the questionnaire to assess the least pain in the past 3 months, on a scale from 1 to 10. Also included in the questionnaire was the Short-Form McGill Pain Questionnaire (SF-MPQ), which consists of 15 pain descriptors rated by the respondent as 0 (none), 1 (mild), 2 (moderate), and 3 (severe). 28 The first 11 descriptors represent the sensory dimension, and the other 4 represent the affective dimension. The range of scores for the sensory component is 0 to 33, and the range of scores for the affective component is 0 to 12. Subjects were asked to complete the SF-MPQ for each pain problem. The SF-MPQ correlates highly with the sensory, affective, and total scores of the original MPQ and is sensitive to the effects of treatments for pain. 28 We added 8 descriptors (stinging, cutting, piercing, radiating, tight, nagging, squeezing, tingling) from the original MPQ that were found by Davidoff et al 29 to be used frequently to describe pain associated with SCI, as well as 1 additional descriptor, shocking. These words were included for descriptive purposes and were not used in the scoring of the SF-MPQ. Procedures Pain categorization was performed by first comparing the pain location and distribution with the subject s level of injury. This information was combined with a classification of chronic SCI pain that uses a matrix that compares the type of pain with location and with the effects of activity, position, and light touch on pain (table 2). 30 If the pain did not seem to fit a specific category on the basis of the information, the subject s self-reported source of pain (musculoskeletal or nervous system) and pain exacerbators were used to help make the final Pain Category (major) Pain Category (specific) Table 2: Criteria for Categories of Pain Location Related to Activity Affected by Position Worse With Light Touch Neurologic SCI pain Below injury in area without normal sensation Transition zone At level of injury, bilateral pain Radicular pain At any dermatomal level, usually unilateral, usually radiates Visceral pain In abdomen Musculoskeletal Mechanical In back or neck, often bilateral spine pain Overuse pain Often above injury in areas of normal sensation, in an incomplete injury can be below Abbreviations:, yes;, no;, maybe. Reprinted with permission. 30

4 PAIN CLASSIFICATION, Cardenas 1711 Table 3: Pain Intensity and Pain-Related Disability for Worst Pain Type of Pain Characteristic Pain Intensity Pain-Related Disability Least Pain* SCI pain (n 52 [31.9%]) Radicular pain (n 15 [9.2%]) Transition zone (n 9 [5.5%]) Visceral pain (n 5 [3.1%]) Mechanical spine pain (n 46 [28.2%]) Overuse pain (n 15 [9.2%]) Unable to categorize (n 21 [12.9%]) (N 163) NOTE: Values are mean SD. * Least pain scores multiplied by 10 for ease of comparison. categorization. Pain problems that could not be categorized by the previous criteria were designated as unable to categorize. In this study, we chose the term spinal cord injury pain to describe the diffuse pain located below the level of the lesion; this is the most common form of neuropathic pain after SCI. The literature contains many other terms (central pain, dysesthetic pain, diffuse pain, phantom body pain) for this same phenomenon. All pain problems were categorized by 1 investigator. To determine interrater reliability, 25% of questionnaires were categorized independently by 2 investigators, a nurse practitioner and a physiatrist. They categorized 41 subjects with 68 pain problems. Subjects were also invited to participate in a double-blind, randomized, controlled trial to evaluate the efficacy of amitriptyline as compared with a placebo in relieving chronic pain associated with SCI. Results of the drug study, in which 84 subjects participated, were recently published, 31 and, as a part of the study, subjects were interviewed in person to classify pain type. Fifteen of the 84 subjects were interviewed by the same 2 investigators to establish the interrater reliability of the categorization of their pain. They were evaluated separately according to the following steps: (1) review pain drawings from the survey questionnaire and the subject s priority numbering system (1 worst pain, 2 second worst pain); (2) determine the subject s level of injury on the basis of his/her response to the questionnaire; (3) for each pain, determine whether it was above, at, or below the level of injury, on the basis of the pain location and the level of injury; (4) for each pain, determine whether it was unilateral or bilateral; (5) for each pain, determine what made it better or worse through answers to questions related to the effect of activity, position, and light touch; and (6) use Criteria for Categories of Pain (see table 2) to classify each pain location according to the best fit for pain category. The statistic 32 was used as the measure of interrater reliability. The amount of agreement beyond chance based on the value is defined as less than 0.0, poor;.00 to.20, slight;.21 to.40, fair;.41 to.60, moderate;.61 to.80, substantial; and.81 to 1.00, almost perfect. 33 RESULTS Interrater Reliability Of the questionnaires categorized independently by 2 investigators, strength of agreement was substantial in categorizing the 68 pain problems (.68). In this group of 41 subjects, the most common worst pain was SCI pain (30.9%), followed by mechanical spine pain (27.9%), overuse pain (13.2%), transition zone pain (7.4%), radicular pain (5.9%), visceral pain (2.9%), and unable to categorize (11.8%). These proportions are comparable to those of the worst pain for the total group with pain (table 3). Of 15 persons whose pain was categorized in person, 13 had 3 pains each, and 2 had 2 pains each, for a total of 43 pain problems. Strength of agreement between the investigators was substantial (.66). The most common worst pain was SCI pain (37.2%), followed by mechanical spine pain (25.6%), overuse pain (20.9%), radicular pain (7.0%), visceral pain (4.7%), and transition zone pain (4.7%). There were no pains that could not be categorized. Pain Categorization Of the 163 respondents who reported pain, 124 (57.5%) reported more than 1 pain. The median duration of Pain 1 (worst pain) was 3.41 years and of Pain 2 was 2.97 years. The duration of worst pain ranged from a minimum of 4.68 months to a maximum of 29.9 years. The categorization of the worst pain (Pain 1) classified on the basis of the questionnaire is shown in table 3. Pain could not be categorized in 12.9% of the respondents. Pain 2 was categorized as shown in table 4; 12.9% of the second pains could not be categorized. For 92 (56.4%) persons, either the worst or second worst pain was categorized as SCI pain. The next most common pain was mechanical spine pain, which was identified in 66 (40.5%) persons. Among those with 2 pains, the most common pain combination was SCI pain and mechanical spine pain (n 23). Pain Intensity and Interference With Activities On average, for worst pain, respondents reported a high level of pain intensity (mean characteristic pain intensity standard Table 4: Pain Intensity and Pain-Related Disability for Second Worst Pain Type of Pain Characteristic Pain Intensity Pain-Related Disability Least Pain* SCI pain (n 39 [32.3%]) Radicular pain (n 6 [4.8%]) Transition zone pain (n 11 [8.9%]) Visceral pain (n 5 [4.8%]) Mechanical spine pain (n 19 [16.1%]) Overuse pain (n 25 [20.2%]) Unable to categorize (n 16 [12.9%]) (N 121) NOTE: Values are mean SD. * Least pain scores multiplied by 10 for ease of comparison. Three subjects not included in table due to incomplete data.

5 1712 PAIN CLASSIFICATION, Cardenas deviation, ; scale range, 0 100) and moderate painrelated disability (avg of 3 pain interference with activities ratings, ; scale range, 0 100). Characteristic pain intensity and pain-related disability scores are shown in table 3 for the worst pain and in table 4 for the second worst pain. Characteristic pain intensity and least pain intensity were highest for respondents whose worst pain could not be classified (table 3). Pain-related disability was also high in this group. Pain-related disability scores for worst pain were 40 in 91% of respondents and 50 in approximately 18%. Analysis with 1-way analysis of variance (ANOVA) indicated a statistically significant difference among the groups of worst pain categories in characteristic pain intensity scores (F 3.566, P.008) and a trend toward a significant difference in pain-related disability (F 2.352, P.057). Further analysis with the Tukey pairwise comparison procedure showed that for those whose worst pain could not be categorized, characteristic pain intensity was significantly greater than for those whose worst pain was mechanical spine pain (P.008) or overuse pain (P.016). No significant differences in pain intensity or pain-related disability were found among pain categories for second worst pain. Qualitative Pain Descriptors The SF-MPQ words most frequently used to describe SCI pain (chosen by more than half of respondents) were hotburning (65.4%), aching (63.5%), tiring-exhausting (59.6%), cramping (57.7%), sharp (53.8%), and throbbing (51.9%). In addition, more than half of those with SCI pain endorsed several words contained in the original MPQ, but not in the SF-MPQ: tingling (71.2%), tight (57.7%), stinging (53.8%), radiating (51.9%), and nagging (51.9%). The SF-MPQ words most frequently used to describe mechanical spine pain (chosen by more than half of respondents) were aching (93.5%), tiring-exhausting (76.1%), sharp (69.6%), tender (58.7%), stabbing (56.5%), shooting (54.3%), and throbbing (52.2%). In addition, more than half of the subjects with mechanical pain endorsed 2 words contained in the original MPQ but not the SF-MPQ: nagging (76.1%) and tight (71.7%). The SF-MPQ words most frequently used to describe transition zone pain (chosen by more than half of respondents) were shooting (55.6%), stabbing (55.6%), sharp (66.7%), hotburning (55.6%), and aching (66.7%). In addition, more than half of subjects with transition zone pain endorsed 3 words contained in the original MPQ but not in the SF-MPQ: piercing (66.7%), tight (55.6%), and nagging (55.6%). There were 21 respondents for whom worst pain could not be categorized. More than half of these respondents endorsed 58% of all descriptors of pain in our measure. The SF-MPQ words most frequently chosen were throbbing (71.4%), hotburning (71.4%), aching (71.4%), tiring-exhausting (71.4%), heavy (66.7%), cramping (66.7%), sharp (61.9%), stabbing (61.9%), and tender (57.1%). The percentages of respondents who endorsed the 5 words contained in the original MPQ were tingling (90.5%), nagging (76.2%), stinging (71.4%), tight (66.7%), and radiating (61.9%). Statistical analyses were not conducted to compare pain categories on each word because of the fairly small values and the large number of tests this would entail, thus risking spurious results. However, data are presented for descriptive purposes and to suggest potentially fruitful areas for future research. Table 5 shows respondents SF-MPQ total, affective, and sensory scores for worst pain. ANOVAs showed no statistically significant difference among the different pain categories Table 5: Pain Descriptors for Worst Pain SF-MPQ Scale SCI pain (n 52) Radicular pain (n 15) Transition zone pain (n 9) Visceral pain (n 5) Mechanical spine pain (n 46) Overuse pain (n 15) Unable to categorize (n 21) Mean SD on the SF-MPQ total and affective scale scores, but there was a trend toward a statistically significant difference on the sensory subscale (P.06). Post hoc comparisons showed that sensory scale scores were significantly higher in the group whose worst pain could not be categorized (mean, 15.57) than in the overuse pain group (mean, 8.87, P.03). CPG Rating The categories of pain were combined into neurogenic (SCI pain, radicular pain, transition zone pain, visceral pain), musculoskeletal (mechanical spine pain, overuse pain), and unable to categorize in order to facilitate their comparison in terms of the CPG categories. The percentage of persons within each of the 4 CPG categories for those with neurogenic, musculoskeletal, and unable to categorize pains is shown in figure 1. The proportions with severe pain-related disability (grade IV) differed across pain categories. Of the 61 respondents with musculoskeletal pain, 26.2% were grade IV; of the 81 respondents with neuropathic pain, 29.9% were grade IV. Of the 21 respondents whose pain could not be categorized, 60% were grade IV. The percentage of persons within each of the 4 categories of the CPG for each pain category for worst pain is shown in table 6. DISCUSSION Using a classification system with 6 major categories, we show substantial interrater reliability in categorizing types of pain problems in persons with SCI. The use of a matrix with information on level of injury and on completeness of injury helped with the categorization; however, of the 68 pains categorized with the questionnaires, there were 8 (11.8%) that could not be categorized on the basis of survey questions alone. In the group of 15 subjects who were interviewed in person, all

6 PAIN CLASSIFICATION, Cardenas 1713 Fig 1. CPG: percentage of each grade within each worst pain category. pains were categorized. Interviewing the subjects clearly added a component not available through the questionnaire alone. The subjects were able to elaborate on their answers to questions, thus providing the interviewer with a better description of the pain problem. The results of this study are consistent with those of other recent studies that found significant levels of pain intensity and pain-related disability in people with SCI. Our results extend those findings by providing preliminary evidence that ratings of pain intensity and pain-related disability may differ according to the type of pain. However, the relatively small number within each category of pain limited our ability to conduct and draw reliable conclusions from statistical comparisons of categories. Further research should examine the influence of type of pain on pain-related disability and on quality of life. The taxonomy we used differs from that of a recent International Association for the Study of Pain (IASP) publication 13 in at least 3 ways. The first level of the IASP classification divides pain into nociceptive and neuropathic. Our classification uses the terms neurologic and musculoskeletal. Second, the IASP taxonomy has a second grouping level that further divides nociceptive pain into musculoskeletal and visceral pain types. Neuropathic pain is divided into above-level, at-level, and below-level pain. In our taxonomy, the location of pain (above, at, and below level) is used as a descriptor for each of the 6 major pain categories. Third, unlike the IASP, but similar to other investigators, 5,21 our taxonomy classifies visceral pain as a type of neuropathic pain. Our experience is that some persons with SCI have unexplained chronic abdominal pain in the absence of demonstrable visceral pathology. The pain found in a bandlike area at the level of the lesion has been described as segmental, segmental deafferentation, or transitional zone pain. It has been defined as pain located within 2 dermatomes above or below the neurologic level. 12 Some investigators have combined transition zone pain with SCI pain when determining prevalence 34 ; however, the pathologic processes are different, and transition zone pain is often associated with allodynia and hyperalgesia. We used the term transitional zone pain but did not require location within 2 dermatomes because we were using pain drawings to determine location and distribution along with the selection of at-level pain. The term musculoskeletal pain is also used in different ways to describe back pain in SCI. Some investigators state that mechanical and musculoskeletal pain below the level of the lesion cannot occur in complete injuries or unless the zone of partial preservation includes the level of the pain. 11,12 However, Frisbie and Aguilera 35 reported that 11 of 19 people with musculoskeletal pain had complete SCI. Possible explanations are that the innervation of deeper structures in the back does not necessarily conform to the innervation of the skin and that lesions that seem complete still allow for pain transmission. 5 We used the term mechanical spine pain to describe localized back or neck pain without radiation, but we did not determine whether there was any spinal instability, which we believed to be unlikely because injuries were not acute. We relied on the association with activity, location of pain on the basis of the pain drawing, and lack of allodynia for classification of mechanical spine pain and did not consider completeness of the lesion to be a factor. We found a high average characteristic pain intensity score (61.02) for the total group of respondents with pain, despite ongoing treatment in most. Those categorized as having overuse pain as worst pain had the lowest average pain-related disability score. A greater proportion of persons with overuse pain had a CPG of grade I (low pain and low disability) as compared with other pain categories. Musculoskeletal pain in often described as an aching pain. 13,35 SCI pain is typically described as burning or stinging. 5,21,35 The SF-MPQ was used in this study to describe each type of pain. Note that aching, a term used by 93.5% of those whose worst pain was categorized as mechanical spine pain and by 80% of those with overuse pain syndromes, was also used by 63.5% of persons with SCI pain to describe their pain and by 66.7% to describe transition zone pain. In this sample, at least, aching was not a term that ruled out neuropathic pain. The terms hot-burning and tingling were used much more frequently to describe pain categorized as SCI pain than to describe mechanical spine and overuse pain. Sixty-five percent of SCI pain problems were described as hot-burning, whereas only 39.1% of mechanical spine pain and 33.3% of overuse pain syndromes were described as hot-burning. Five words in the original MPQ, but not the SF-MPQ (tingling, tight, stinging, radiating, nagging), were chosen by more than half of those with SCI pain to describe this pain. Further research is needed to compare the original MPQ with the SF-MPQ in the SCI population, because it is possible that the SF-MPQ may fail to capture fully the range of qualitative descriptors of SCI pain. We did not use the SF-MPQ responses to categorize pain and, therefore, do not know whether incorporation of the SF-MPQ descriptors might have altered the reliability that we found. This study has several methodologic limitations. First, only 60% of the mailed questionnaires were returned, excluding Table 6: Distribution of CPG Within Each Pain Category for Worst Pain Type of Pain Grade I Grade II Grade III Grade IV SCI pain (n 44) Radicular pain (n 15) Transition zone (n 9) Visceral pain (n 5) Mechanical spine pain (n 46) Overuse pain (n 15) Unable to categorize (n 20)

7 1714 PAIN CLASSIFICATION, Cardenas those mailed to addressees later determined to be deceased, ineligible for the study, or not living at that address. Second, the small number of subjects in some pain categories limited our ability to conduct statistical analyses comparing different groups. Third, although there was substantial interrater agreement, we do not know the validity of our categorization method. We did not perform diagnostic tests to rule out other causes of pain, such as syringomyelia, or to confirm diagnoses. Fourth, we asked subjects to determine their worst pain and cannot know whether their choices would be different if they were asked at a different point in time. Finally, although the sample is fairly representative of individuals in the NWRSCIS, respondents may differ in sociodemographic and other characteristics from the larger population of SCI individuals in the United States. Thus, the proportions of pain categories found in this study may or may not be similar to those in the general population of adults with SCI. Despite these limitations, this study is the first to report an SCI pain classification scheme with proven interrater reliability. In addition, it provides some intriguing findings that need to be replicated and extended in future studies. CONCLUSIONS By using responses to specific questions, level of injury, and a pain drawing, we achieved substantial interrater reliability in determining pain categories with a classification system based on presumed pathology. Adding interviews with patients increased our ability to classify pain, although it did not alter overall interrater reliability. Our survey indicates that most persons with SCI experience chronic pain. Although certain qualitative descriptors of pain are more often associated with a specific type of pain, descriptors are not pathognomonic for a specific type of pain. More research is needed to increase our understanding of the use of descriptors in this population. The development of a reliable classification system is an important step toward improving our ability to compare future findings from treatment studies. References 1. Michaelis LS. The problem of pain in paraplegia and tetraplegia. Bull N Y Acad Med 1970;46: Guttmann L. Disturbances of sensibility. In: Guttman L, editor. Spinal cord injuries: comprehensive management and research. 2nd ed. Oxford: Blackwell Scientific; p Burke DC, Woodward JM. Pain and phantom sensation in spinal paralysis. In: Vinken PJ, Bruyn GW, editors. Handbook of clinical neurology. Vol 26. New York: Elsevier; p Donovan WH, Dimitrijevic MR, Dahm L, Dimitrijevic M. Neurophysiological approaches to chronic pain following spinal cord injury. Paraplegia 1982;20: Loubser PG, Donovan WH. Chronic pain associated with spinal cord injury. In: Narayan RK, Wibberger JE, Povlishock JT, editors. Neurotrauma. New York: McGraw-Hill; p Mariano AJ. Chronic pain and spinal cord injury. Clin J Pain 1992;8: Britell CW, Mariano AJ. Chronic pain in spinal cord injury. Phys Med Rehabil 1991;5: Rose M, Robinson JE, Ellis P, Cole JD. Pain following spinal cord injury: results from a postal survey. Pain 1988;34: Demirel G, Yllmaz H, Gencosmanoglue B, Kesiktas N. Pain following spinal cord injury. Spinal Cord 1998;36: Merskey H, Bogduk N. Task force on taxonomy of the International Association for the Study of Pain. Classification of chronic pain: descriptions of chronic pain syndromes and definitions of pain terms. 2nd ed. Seattle: IASP Pr; p Siddall PJ, Taylor DA, Cousins MJ. Classification of pain following spinal cord injury. Spinal Cord 1997;35: Bryce TN, Ragnarsson KT. Pain after spinal cord injury. Phys Med Rehabil Clin North Am 2000;11: Siddall PJ, Yezierski RP, Loeser JD. Pain following spinal cord injury: clinical features, prevalence, and taxonomy. IASP Newsl [serial online]. 2000;3. Available at: TC00-3.html. Accessed May 15, Turner JA, Cardenas DD. Chronic pain problems in individuals with spinal cord injuries. Semin Clin Neuropsychol 1999;4: Davidoff G, Guarracini M, Roth E, Sliwa J, Yarkony G. Trazodone hydrochloride in the treatment of dysesthetic pain in traumatic myelopathy: a randomized, double-blind, placebo-controlled study. Pain 1987;29: Cairns DM, Adkins RH, Scott MD. Pain and depression in acute traumatic spinal cord injury: origins of chronic problematic pain? Arch Phys Med Rehabil 1996;77: Botterell EH, Callaghan JC, Jousse AT. Pain in paraplegia: clinical management and surgical treatment. Proc R Soc Med 1953;47: Woolsey RM. Chronic pain following spinal cord injury. J Am Paraplegia Soc 1986;9: Davis R, Lentini R. Transcutaneous nerve stimulation for treatment of pain in patients with spinal cord injury. Surg Neurol 1975;4: Fenollosa P, Pallares J, Cervera J, Pelegrin F, Inigo V, Giner M. Chronic pain in the spinal cord injured: statistical approach and pharmacological treatment. Paraplegia 1993;31: Anke AG, Stenehjem AE, Stanghelle JK. Pain and life quality within two years of spinal cord injury. Paraplegia 1995;33: Summers JD, Rapoff MA, Varghese G, Porter K, Palmer RE. Psychosocial factors in chronic spinal cord injury pain. Pain 1991;47: von Korff M, Ormel J, Keefe FJ, Dworkin SF. Grading the severity of chronic pain. Pain 1992;50: Smith BH, Penny KI, Purves AM, Munro C, Wilson B, Grimshaw J. The chronic pain questionnaire: validation and reliability in postal research. Pain 1997;71: Underwood MR, Barnett AG, Vickers MR. Evaluation of two time-specific back pain outcome measures. Spine 1999;24: Penny KI, Purves AM, Smith BH, Chambers WA, Smith WC. Relationship between the chronic pain grade and measures of physical, social and psychological well-being. Pain 1999;79: Dworkin SF, von Korff M, Whitney CW, Le Resche L, Bicker BG, Barlow W. Measurement of characteristic pain intensity in field research [abstract]. Pain Suppl 1990;5:S Melzack R. The short-form McGill Pain Questionnaire. Pain 1987;30: Davidoff G, Roth E, Guarracin M, Sliwa J, Yarkony G. Function-limiting dysesthetic pain syndrome among traumatic spinal cord injury patients: a cross-sectional study. Pain 1987;29: Cardenas D. Current concepts of rehabilitation of spinal cord injury patients. Spine State Art Rev 1999;13: Cardenas DD, Warms CA, Turner JA, Marshall H, Brooke MM, Loeser JD. Efficacy of amitriptyline for relief of pain in spinal cord injury: results of a randomized controlled trial. Pain 2002; 96: Fleiss JL. The measurement of interrater agreement. In: Fleiss JL, editor. Statistical methods for rates and proportions. New York: John Wiley & Sons; p Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33: Stormer S, Gerner HJ, Gruninger W, Metzmacher K, Follinger S, Wienke C. Chronic pain/dysaesthesiae in spinal cord injury patients: results of a multicentre study. Spinal Cord 1997;35: Frisbie JH, Aguilera EJ. Chronic pain after spinal cord injury: an expedient diagnostic approach. Paraplegia 1990;28:460-5.