 |
|
|
The Agency for Healthcare Research and Quality (AHRQ), through its Evidence-based Practice Centers (EPCs), sponsors the development of evidence reports and technology assessments to assist public- and private-sector organizations in their efforts to improve the quality of health care in the United States. The reports and assessments provide organizations with comprehensive, science-based information on common, costly medical conditions and new health care technologies. The EPCs systematically review the relevant scientific literature on topics assigned to them by AHRQ and conduct additional analyses when appropriate prior to developing their reports and assessments.
To bring the broadest range of experts into the development of evidence reports and health technology assessments, AHRQ encourages the EPCs to form partnerships and enter into collaborations with other medical and research organizations. The EPCs work with these partner organizations to ensure that the evidence reports and technology assessments they produce will become building blocks for health care quality improvement projects throughout the Nation. The reports undergo peer review prior to their release.
AHRQ expects that the EPC evidence reports and technology assessments will inform individual health plans, providers, and purchasers as well as the health care system as a whole by providing important information to help improve health care quality.
We welcome written comments on this evidence report. They may be sent to: Director, Center for Practice and Technology Assessment, Agency for Healthcare Research and Quality, 6010 Executive Blvd., Suite 300, Rockville, MD 20852.
| Carolyn Clancy, M.D. | Robert Graham, M.D. |
| Acting Director | Director, Center for Practice and Technology |
| Agency for Healthcare Research | Assessment |
| and Quality | Agency for Healthcare Research and Quality |
| The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services of a particular drug, device, test, treatment, or other clinical service. |
To evaluate (a) the measurement of central neuropathic pain (CNP) after traumatic spinal cord injury (TSCI), (b) the prevalence of acute and chronic CNP, (c) predictive factors for chronic CNP, and (d) the effectiveness and safety of various interventions for CNP.
Studies were identified by searching MEDLINE, EMBASE, and PsycINFO (to May 2000); CINAHL, HEALTHStar, and Sociological Abstracts (to November 1999); the Cochrane Library (issue 4, 1999); reference lists of eligible articles found in the searches; and personal files of advisory panel members.
Studies about the cause, management, or measurement of CNP were included. Exclusion criteria were participants without TSCI or CNP, children younger than 13 years, or inability to determine whether chronic pain was central and neuropathic.
Two reviewers independently extracted data for all studies except case reports. One reviewer extracted case report data, which was checked by another. Disagreements were resolved by consensus. The quality of data was assessed. Data were not pooled because clinical heterogeneity existed across studies, outcome measurements were inconsistent, studies had low methodological quality, and data reporting was incomplete.
132 unique studies (6 randomized controlled trials and 126 observational studies, including 47 case series and at least 56 single or multiple case reports) met the selection criteria. Few studies evaluated the management of CNP following TSCI in women; and no studies evaluated adolescents only, the role of treatment algorithms, or multidisciplinary approaches. Only two studies evaluated self-management strategies in cases of CNP following TSCI. Diagnosis, assessment and natural history: No discriminative or evaluative measurement instruments have been adequately investigated with respect to psychometric measurement properties in this setting. Pharmacological interventions: Little research has been done. The few studies available have such poor methods that it was not possible to evaluate interventions. It appears that local anesthetics, opioids, and clonidine given spinally may be effective in relieving CNP following TSCI, but better research is needed. Spinal cord and deep brain stimulation techniques: The studies had similar deficiencies to those described above. The limited evidence available suggests that spinal cord stimulation has a variable rate of early success and a low rate of long-term effectiveness. Deep brain stimulation has a low rate of early success and an even lower long-term success rate, coupled with significant adverse events. Dorsal root entry zone (DREZ) lesions and other surgical interventions: All studies on DREZ showed high rates of success but had poorly defined or no eligibility criteria, included no control groups, and did not report adequately the severity of adverse effects.
This report describes rigorous systematic reviews on the measurement and management of CNP after TSCI in adults and adolescents. The research on this topic is in its infancy. This report describes the limitations of the available evidence and provides recommendations for future research.
This document is in the public domain and may be used and reprinted without permission except those copyrighted materials noted for which further reproduction is prohibited without the specific permission of copyright holders.
Jadad A, O'Brien MA, Wingerchuk D, et al. Management of Chronic Central Neuropathic Pain Following Traumatic Spinal Cord Injury. Evidence Report/Technology Assessment Number 45. (Prepared by McMaster University Evidence-based Practice Center under Contract No. 290-97-0017.) AHRQ Publication No. 01-E063. Rockville, MD: Agency for Healthcare Research and Quality. September 2001.
Pain has been recognized for more than half a century as one of the many symptoms experienced by people who have suffered traumatic spinal cord injuries (TSCI). Efforts to estimate the prevalence, severity, and duration of pain after TSCI have led to variable results. This variation has been explained by differences among the studies in terms of pain definitions, terminology, classification, inclusion criteria, variability in reporting methods, as well as several etiological, demographic, and cultural factors.
Great variability and little consensus have also plagued the classification of pain in people with TSCI. In 1997, however, a group of investigators developed a classification that seems to be gaining widespread acceptance. The first axis of this classification includes four major categories or divisions of pain: "musculoskeletal," "visceral," "neuropathic," and "other." These categories are based on the system affected, which can be readily identifiable in clinical settings. Neuropathic pain is the focus of this report.
Neuropathic pain is defined as pain that occurs following damage to the central or peripheral nervous system. This pain can be identified by site (region of sensory disturbance) and by features (sharp, shooting, electric, burning, stabbing). Neuropathic pain can be further broken down by site (Axis 2) into neuropathic pain "at level" (pain that occurs at the level of the SCI, in a segmental pattern with neuropathic features) and neuropathic pain "below level" (diffuse pain that is described by the words "burning," "tingling," "aching," "shooting," or "stabbing" and that should be present at least three segments below the level of injury). Neuropathic pain "at level" can be subdivided further into radicular (when it can be attributed to nerve root pathology) or central (when it is due to changes within the spinal cord or possible supraspinal structures), although the value of this subdivision has been questioned.
Neuropathic pain not only is one of the most challenging conditions in chronic pain management and one of the most promising areas in pain research, but it also may have even greater impact on the quality of life of patients than the extent of the injury itself. The objectives of this Task Order were to conduct a comprehensive, systematic review of the literature on this important topic and to support guideline development initiatives by the Consortium for Spinal Cord Medicine (CSCM) and other interested organizations while building on existing work and focusing on answerable, clinically relevant questions.
A set of questions was initially proposed by the CSCM and further refined with input from members of the McMaster University Evidence-based Practice Center (MU-EPC) and the Task Order Officer (TOO). All questions, unless otherwise specified, relate to the assessment or management of chronic central neuropathic pain (CNP) following TSCI in adults and adolescents.
After multiple consultations, the following questions were selected as the focus of the evidence report. To maximize the efficiency of the process, they were grouped by theme.
What are the measurement properties (reliability, validity, and responsiveness) of:
assessment approaches for chronic CNP per se (including criteria and tools such as inventories, questionnaires, and scales);
other outcome measures or assessments (related to the experience of pain); and
assessment approaches (including criteria and tools such as inventories, questionnaires, and scales) to identify new-onset musculoskeletal pain against a background of chronic CNP?
What is the strength of evidence for strategies for the differential diagnosis of chronic CNP from other types of pain?
What is the strength of evidence for identifying the prevalence of acute and chronic CNP and factors that could predict the development of chronic CNP?
a) What is the evidence for the effectiveness and safety of each of the following classes of medications: simple analgesics (including nonsteroidal anti-inflammatory drugs [NSAIDs] and acetaminophen); antidepressants (including tricyclics and selective seratonin reuptake inhibitors [SSRIs]); antiseizure medication; narcotics; muscle relaxants; N-methyl-D-aspartate (NMDA) antagonists; and local anesthetics?
b) How do these classes of medications compare with one another?
c) What is the strength of evidence for the effectiveness and safety of treatment algorithms including these classes of medication?
What is the evidence of the effectiveness and safety of: (a) transcutaneous electrical nerve stimulation (TENS); (b) nerve blocks (regional anesthetic interventions); (c) surgery, including dorsal root entry zone (DREZ); (d) multidisciplinary pain treatment approaches;(e) pain management approaches; (f) comprehensive pain management clinics; and (g) psychosocial interventions?
What is the evidence for the effectiveness and safety of self-management approaches to chronic pain management (e.g., Catalano's workbook, Caudill's workbook, Aspen's pain management education manual, Tollison's pain management patient guide)?
What are the costs (if available) associated with pharmacologic, technological, and other interventions listed in questions 4, 5, and 6 above?
To our knowledge, there have been no systematic reviews specifically designed to answer any of the questions formulated by the CSCM.
The Technical Expert Panel (TEP) for this Task Order included individuals who represented professional organizations, providers of health care, purchasers of health services, researchers, and consumers. These individuals are recognized as national and international leaders in the management of pain or in issues related to spinal cord injury.
Initially, we used very liberal selection criteria. We regarded as potentially eligible any article that described a study: 1) in humans and 2) about the cause, management, or measurement of CNP in individuals after TSCI. There was no exclusion based on study design.
We excluded reports that were not primary studies and studies where the sample consisted of people without a TSCI, those without chronic neuropathic pain, or children younger than 13 years. We also excluded primary studies that did not contain data in the published report and studies in which the sample included individuals with TSCI as well as other types of CNP, but where the results were not presented separately for individuals with TSCI. In addition, we excluded studies that only used the term "chronic pain" without any other description of the pain experienced by the individuals in the study sample that could have helped us judge it as central and neuropathic.
We accepted any definition for CNP provided by the primary study authors. We developed a list of descriptions of CNP based on our preliminary searches. This list was used to guide the research team during the two-step screening process with six raters working in pairs. The first step was based upon the information available in titles and abstracts (where available) and was conducted by the same two raters working independently. The second step of the screening was based upon full text reports and involved all six raters randomly paired. Discrepancies were resolved by discussion.
Citations of potentially eligible studies were identified through a systematic search of:
MEDLINE, EMBASE, and PsycINFO were searched from the date of their release to end of May 2000.
CINAHL, HEALTHStar, and using Sociological Abstracts. All databases were searched from the date of their release to November 1999.
The reference lists of any eligible article identified in any of the above sources.
Personal files of all members of the local team and the TEP.
The development and refinement of the search strategy followed an iterative process using MEDLINE. The refined MEDLINE strategy was modified to meet the specific features of CINAHL, EMBASE and PsycINFO.
All data extraction forms were developed, pilot-tested, and revised by members of the local research team, including the team statisticians. After consultations with the TEP, our TOO, and our partners, the forms were approved for content. A general data extraction form was used with all studies, while individual forms were used for randomized controlled trials (RCTs), observational studies, and case reports. Items related to the quality of different study designs were embedded within the data extraction forms. Two reviewers completed data extraction independently for all studies except the case reports. For these studies, data were extracted by one reviewer and checked by another. Any disagreements were resolved by consensus. Following consensus on each item, the data forms were scanned into a Microsoft Access database using Teleform software.
Descriptive statistics were calculated for all fields of the database. Evidence tables were constructed to describe the most salient features of the included studies according to the review question. These tables are found at the end of each chapter along with the relevant supplementary tables.
The local research team at the MU-EPC, in consultation with members of the partner organizations and the TOO, evaluated the overall quantity and quality of the data available. This evaluation led to the conclusion that meta-analysis would be inappropriate to summarize the evidence on each of the research questions or for each of the main categories of interest. The main reasons for this decision were substantial clinical heterogeneity across the studies (e.g., interventions evaluated, patient samples, duration); inconsistency in outcome measurements; low methodological quality; and incomplete data reporting (see detailed descriptions within each category). Therefore, this report represents a systematic, qualitative review of the existing evidence, emphasizing the implications for clinical practice and the directions that future researchers could take to fill existing knowledge gaps.
For the purposes of this Evidence Report, the evidence syntheses were grouped in five chapters that included:
The yield of the literature and the general characteristics of all the studies included;
Studies on the diagnosis, assessment, and natural history of CNP after TSCI;
Pharmacological interventions;
Spinal cord and deep brain stimulation techniques; and
DREZ lesions and other surgical interventions.
The analysis of the yield of the literature and the general characteristics of the studies showed that:
A total of 591 full articles were retrieved and screened. After screening, 158 studies met the inclusion criteria. Of the 158 studies, 19 were reported in more than one publication. After several iterations, a total of 132 unique studies were included. These form the basis for the Evidence Report.
Six studies were RCTs, and 126 were observational studies, including 47 case series and at least 56 single or multiple case reports.
Overall, numerous deficiencies in the reporting of the studies limited the assessment of their validity, relevance, precision and, therefore, their clinical application. More than 50 percent of studies did not provide a definition for neuropathic pain, report the cause of the injuries, describe the use of surgical stabilization, state the onset time for pain after injury, or highlight the duration of pain. Fifty-four percent of the studies did not report the time from the injury to the inclusion in the study, the completeness of the injury, or the area of the body affected by pain.
There was little information on the management of CNP following TSCI in women and adolescents.
Thirty percent of studies had fewer than 25 patients. This limited their power to detect meaningful, clinically important differences among the interventions.
There were no studies that evaluated the role of treatment algorithms or multidisciplinary approaches. Only two studies evaluated self-management strategies in cases of CNP following TSCI.
Comparison or synthesis of data across studies was limited by the low quality of reporting and by the large number and heterogeneity of outcome measures and tests used in the studies.
The following is a description of the main conclusions and the implications from practice that could be derived from the evidence available to address the initial questions:
Diagnosis, assessment, and natural history: There are no discriminative or evaluative measurement instruments that have been adequately investigated with respect to psychometric measurement properties in this context. Despite the serious limitations of most of the individual studies, most estimates of the prevalence of chronic pain after TSCI vary from 40 percent to 75 percent of patients. Pain is moderate to severe in 25 percent to 60 percent of these individuals, is often associated with psychological and psychiatric conditions, and is severe enough to impair or prevent optimal physical function and daily living.
Pharmacological interventions: There is a dearth of research in this area, which includes most of the interventions that are regarded as the core for the management of other types of neuropathic pain. The few studies available have such small sample sizes, poor methodology, and incomplete reports that it was not possible to judge the value of any individual intervention or group of interventions. Although it appears that local anesthetics, opioids, and clonidine given spinally may be effective in relieving CNP following TSCI, better research is needed. While the needed evidence is gathered directly from patients with CNP after TSCI, clinicians interested in using pharmacological interventions will have to rely on research on these interventions in other patient populations.
Spinal cord and deep brain stimulation techniques: These studies had similar deficiencies to those described above. The limited evidence available suggests that spinal cord stimulation has a variable rate of early success and a low rate of long-term effectiveness. Deep brain stimulation has a low rate of early success and an even lower long-term success rate, coupled with important adverse events. These findings make it difficult to justify the use of either procedure as a method of treating CNP after TSCI. Transcutaneous electrical nerve stimulation may reduce the sensation of "pain unpleasantness" if patients have positive expectations of treatment effectiveness.
Dorsal root entry zone lesions and other surgical interventions: All studies on DREZ showed high rates of success, but had poorly defined (or lacked) inclusion and exclusion criteria, included no control groups, and did not report adequately the severity of the adverse effects experienced by patients. Even recognizing the problems regarding the validity and generalizability of the studies, some may look to DREZ lesioning or other spinal surgeries as a last resort when other palliative efforts have failed. Given that the studies did not adequately report the severity of the adverse effects experienced by patients, it is unknown whether DREZ lesioning and other spinal surgeries pose unwarranted risks to patients.
Research on the management of central CNP following TSCI is in its infancy. The following are some suggestions for future research:
Multicenter collaboration to set a research agenda. The CSCM may be well-positioned to facilitate this level of collaboration, or alternative strategies may be needed to foster pragmatic working relationships, even among groups that do not have a tradition of cooperation. Such collaborative groups could study the research problems and provide training in clinical research to young investigators.
Given the prevalence and severity of CNP following TSCI and the dearth of research to support any therapeutic strategy, it is imperative to develop effective strategies to improve the number, validity, precision, and relevance of future studies.
Larger studies with more rigorous design, more comprehensive reports, and longer term followup are needed to establish the effectiveness and adverse effects of most of the interventions available. Special emphasis should be placed on gathering evidence on the effects of different interventions in women and adolescents.
Research groups should make efforts to select a core set of validated and clinically relevant outcomes to be measured in all the studies, in addition to any other outcomes of interest to the specific groups of researchers.
More rigorous studies, ideally large, double-blind, multicenter RCTs, are clearly needed to establish the relative effectiveness and safety of different pharmacological interventions. Priority should be given to interventions with established roles for the management of other types of neuropathic pain, such as tricyclic antidepressants, anticonvulsants, local anesthetics, and opioids. Studies designed to judge the added value of these interventions given in combination, through invasive routes (e.g., epidural and intrathecal infusions of opioids and local anesthetics), or using different formulations (e.g., sustained release preparations) should also be a priority.
Since CNP is associated with psychosocial difficulties, other noninvasive approaches such as multidisciplinary or self-management approaches should be developed and evaluated for those with TSCI.
More definitive studies are needed to determine the effectiveness and safety of non-pharmacological interventions. Based on the evidence available, the most promising interventions are spinal cord stimulation and DREZ lesions. These interventions, however, are also invasive and potentially harmful. The studies that are needed, however, will require complex, controlled designs with close attention to safety issues, substantial resources, and efficient collaboration among research groups.
Studies are also needed to determine whether the response to treatment is influenced by the level and cause of the SCI; as well as by the duration, distribution, and characteristics of the pain; and by comorbid factors (e.g., anxiety and depressive disorders).
There is a great opportunity for consumer groups to call for and support more research activities, given the number of important questions that remain unanswered.
Funding and conducting the research that is required will not be easy, given the complexity of the disorder, the frequent presence of comorbidity, and the variety of interventions and outcomes available. Future research efforts will require commitment among different groups of stakeholders, some of which do not have a tradition of collaboration.
In summary, this report includes the first set of systematic reviews on the management of chronic CNP following TSCI. They incorporate state-of-the-art methodology and are ready for incorporation into evidence-based clinical practice guidelines or performance measures. The report also provides a detailed description of the many limitations of the evidence available and provides recommendations to fill existing knowledge gaps through rigorous research. Filling such gaps will not be easy and will require highly innovative efforts and collaboration among different groups of decisionmakers. If this field continues to produce few, small, incompletely reported studies with heterogeneous designs instead of the high quality collaborative efforts required, research in this area will continue to be of little value to guide important clinical and policy decisions.
Pain has been recognized for more than half a century as one of the many symptoms experienced by people who have suffered spinal cord injuries (SCI) (Backonja and Galer, 1998; Bedbrook, 1981; Beric, Dimitrijevic, and Lindblom, 1988; Botterell, Callaghan, and Jousse, 1953; Britell and Mariano, 1991; Burke, 1976; Davidoff, Roth, Guarracini, et al., 1987; Davis and Martin, 1947; Davis, 1975; Kaplan, Grynbaum, Lloyd, et al., 1962; Kennedy, 1946; Levi, Hultling, Nash, et al., 1995; Melzack and Loeser, 1978; Michaelis, 1970; Nashold, 1991; Pollock, 1951; Riddoch, 1938; Roth, 1994; Tunks, 1986). Estimates of the prevalence of pain in this group, however, have varied widely. Studies have reported that pain is experienced by 1 to 94 percent of people at some point after suffering an SCI (Ravenscroft, Ahmed, and Burnside, 1999; Roth, 1994; Siddall, Taylor, and Cousins, 1997). Attempts to estimate the severity and duration of pain after SCI have also led to variable results. It has been reported, for instance, that pain can become chronic in 1 to 70 percent and can be severe or disabling in 5 to 37 percent of people after SCI (Kaplan, Grynbaum, Lloyd, et al., 1962; Nepomuceno, Fine, Richards, et al., 1979; Ravenscroft, Ahmed, and Burnside, 1999). This variation in estimates of prevalence, severity, and duration, which has also been shown across institutions within multicenter studies, has been explained by differences among the studies in terms of pain definitions, terminology, classification, inclusion criteria, variability in reporting methods, as well as several etiological, demographic, and cultural factors (Roth, 1994).
Independently of the problems to estimate its prevalence, severity, and duration, it has been shown that pain can be very disabling after SCI. A study has revealed that, depending on the level of the lesion, 23 to 37 percent of people with pain after SCI would, if they had the chance, trade pain relief for loss of bladder, bowel, or sexual function (Nepomuceno, Fine, Richards, et al., 1979).
Studies in animal models have suggested a variety of pathophysiological mechanisms that occur after SCI, from intraspinal sprouting of C, A delta, and A beta primary afferent fibers to alterations in ion channels and transmitter receptor activation state leading to permanent central sensitization of cells in the pain pathway (spinothalamic tract neurons) (Bennett, 1993). Yezierski (1996) reviewed the results of three experimental models of SCI (ischemic, mechanical, and excitotoxic) and concluded that "there are neurochemical, anatomical, and physiological changes that collectively constitute a central injury cascade responsible for the development of clinical symptoms" (Yezierski, 1996). He suggests that it is unlikely that any one mechanism is solely responsible for the onset of central pain following SCI. The clinical field is well positioned to learn from the mammalian studies and model therapeutic interventions based on mechanisms that are known to occur. In the clinic, however, the classification of pain in people with SCI has been plagued by great variability and little consensus.
In 1997, a group of investigators developed a method of classification that seems to be gaining widespread acceptance, as it was built upon the existing literature and designed to provide simple guidance to research and management decisions (Siddall, Taylor, and Cousins, 1997). We will follow this classification, as much as possible, throughout this evidence report.
The first axis of the classification (Axis 1) includes four major categories or divisions of pain: "musculoskeletal," "visceral," "neuropathic," and "other." These categories are based on the system affected, which can be readily identifiable in clinical settings. The following is almost a verbatim description of each of these categories, as stated in the original article (Siddall, Taylor, and Cousins, 1997).
This is pain that arises from damage or overuse in structures such as bones, ligaments, muscles, intervertebral discs, and facet joints. This category also includes mechanical pain due to damage of spinal structures such as that occurring before spinal stabilizing surgery. Musculoskeletal pain can be identified by location (at or above lesion level in those with complete spinal cord lesions) and by features (dull, aching, worse with activities, eased by rest).
This is pain associated with visceral pathology. It can be identified by location (e.g., abdomen) and features (dull, poorly localized, cramping, related to visceral function or pathology).
This is pain that occurs following damage to the central or peripheral nervous system (Merskey and Bogduk, 1994). This pain can be identified by site (region of sensory disturbance) and by features (sharp, shooting, electric, burning, stabbing). Neuropathic pain can be further broken down by site (Axis 2) into neuropathic pain "at level" and neuropathic pain "below level."
As its name indicates, this is pain that occurs at the level of the SCI, in a segmental pattern with neuropathic features. This type of pain may be attributed to nerve root pathology or changes within the spinal cord or possibly supraspinal structures. Siddall and colleagues (1997) also suggest that pain that is described as "burning," "tingling," "sharp," "aching," or "shooting" in a dermatomal distribution at the level of the lesion, with or without hyperesthesia, should be classified as neuropathic at level pain. "At level" should include two segments above and below the level of SCI, because input from several segments may be disrupted or disturbed following injury at any particular level (Siddall, Taylor, and Cousins, 1997).
This type of pain refers to diffuse pain that is described by the words "burning," "tingling," "aching," "shooting," or "stabbing." In distinction to neuropathic at level pain, this pain should be present at least three segments below the level of injury. This type of pain has been labeled by other classification systems as "central pain" or "deafferentation pain."
Neuropathic pain is the focus of this report. Neuropathic pain not only is one of the most challenging conditions in chronic pain management and one of the most promising areas in pain research (Jadad, 1993), but also, it may have even greater impact on the quality of life of patients than the extent of the injury itself (Westgren and Levi, 1998).
This is a category created to include other specific types of pain that are not included in the categories listed above, such as pains that are a "consequence" of SCI. This group includes pains such as those associated with syringomyelia, compressive mononeuropathies, and reflex sympathetic dystrophy.
This classification system excludes sensations following SCI that are not regarded as pain, such as phantom phenomena.
Pain arguably is the best studied of all symptoms. There is a rich tradition of pain relief research spanning the past 50 years. Each year, the body of literature expands by thousands of new articles published in hundreds of journals and books and, increasingly, on the Internet. During the second half of the 20th century, the number of randomized controlled trials (RCTs) in pain relief doubled every 10 years. In January of 2000, a search of the Cochrane Controlled Trials Register (Cochrane Library, 1999), the largest collection of clinical trials in the world, found that "pain" or "analgesia" were mentioned in more than 23,000 citations of RCTs or almost 9 percent of the total. Not surprisingly, keeping up with existing and new knowledge in pain relief has become a difficult task.
The explosion of information on pain and its management, and the need to use it efficiently, has led to an increasing interest in the use of the principles of evidence-based health care to guide clinical decisions and resource allocation around pain relief.
One of the strongest manifestations of the interest in evidence-based decisionmaking has been the considerable increase in the number of systematic reviews addressing a wide variety of pain relief topics. For instance, systematic reviews have been used to evaluate the evidence on the effectiveness and safety of interventions for chronic pain such as transcutaneous electrical nerve stimulation (McQuay and Moore, 1998), intravenous regional sympathetic blockade (Jadad, Carroll, Glynn, et al., 1995; McQuay and Moore, 1998), opioids (Jadad, 1998); anticonvulsant drugs (McQuay, Carroll, Jadad, et al., 1995), antidepressants (McQuay, Tramer, Nye, et al., 1996), systemic local anesthetic-type drugs (Kalso, Tramer, McQuay, et al., 1998; McQuay and Moore, 1998) and cognitive behavior therapy (Morley, Eccleston, and Williams, 1999).
The impetus for the use of systematic reviews to guide decisions in pain relief has been strengthened by the availability of powerful tools such as the Oxford Pain Database (Jadad, Carroll, Moore, et al., 1996), validated tools to assess the quality of analgesic trials (Jadad, Moore, Carroll, et al., 1996), specialized Web sites and textbooks (McQuay and Moore, 1998; Oxford Pain Web Site, 1999), special interest groups within professional organizations (International Association for the Study of Pain, 1999), and the Cochrane Collaboration (Cochrane PaPas Group, 0).
Despite the increasing amount of collaborative work represented by these efforts, however, no systematic reviews have been identified to address issues specifically related to the management of neuropathic pain after TSCI.
Neuropathic pain is perhaps one of the most promising areas for the development of strong partnerships and for profound breakthroughs in clinical treatment. For these efforts to be efficient, it is essential that all the potential partners count with a rigorous synthesis of the best available evidence from clinical research to provide a common platform for their deliberations.
The McMaster University EPC (MU-EPC) was notified in September 1999 that it was successful in its bid to undertake the development of an evidence report on the "Management of Chronic Central Neuropathic Pain Following Traumatic Spinal Cord Injury." This topic was nominated by the Consortium for Spinal Cord Medicine (CSCM), an organization that includes representation from 18 partners (Appendix A). The CSCM plans to use this report as part of its guideline development process.
The objectives of this Task Order were to conduct a comprehensive systematic review of the literature on this important topic and to support guideline development initiatives by the CSCM, while building upon existing work and focusing on potentially answerable, clinically relevant questions.
The absence of a widely accepted definition for "chronic central neuropathic pain," including specific issues related to the chronicity and the traumatic nature of pain, forced a liberal approach to the questions posed by the CSCM in relation to the selection of studies (otherwise we would have had no articles to review). It was decided that the questions would apply to people who had suffered a direct traumatic injury to the spinal cord, that had pain described as chronic (regardless of the duration), and that fit the classification proposed by Siddall et al., 1997.
All questions were initially formulated by the CSCM and further refined with input from members of the MU-EPC and the AHRQ Task Order Officer (TOO). By request from the CSCM and the TOO, all questions, unless otherwise specified, relate to the assessment or management of chronic central neuropathic pain (CNP) following traumatic spinal cord injury (TSCI) in adults.
What are the measurement properties (reliability, validity, sensitivity to change) of:
assessment approaches for chronic central neuropathic pain per se (including criteria and tools such as inventories, questionnaires, and scales to measure pain intensity or relief);
other outcome measures or assessments (related to the experience of pain, such as impact of pain on mood, sleep, and independence); and
assessment approaches (including criteria and tools such as inventories, questionnaires, and scales) to identify new onset musculoskeletal pain against a background of chronic central neuropathic pain?
What is the strength of evidence for strategies for the differential diagnosis of chronic central neuropathic pain from other types of pain?
What is the strength of evidence supporting strategies to estimate the prevalence of acute and chronic central neuropathic pain and factors that could predict the development of chronic central neuropathic pain?
a) What is the evidence for the effectiveness and safety of each of the following classes of medications: simple analgesics (including nonsteroidal anti-inflammatory drugs [NSAIDs] and acetaminophen); antidepressants (including tricyclics and seratonin reuptake inhibitors [SSRIs]); antiseizure medication; narcotics; muscle relaxants; N-methyl-D-aspartate (NMDA) antagonists; and local anesthetics?
b) How do these classes of medications compare with one another?
c) What is the strength of evidence for the effectiveness and safety of treatment algorithms including these classes of medication?
What is the evidence of effectiveness and safety of: (a) electrical stimulation (TENS); (b) regional anesthetic interventions (nerve blocks); (c) surgery, including dorsal root entry zone (DREZ); (d) multidisciplinary pain treatment approaches; (e) pain management approaches; (f) comprehensive pain management clinics; and (g) psychosocial interventions?
What is the evidence for the effectiveness and safety of self-management approaches to chronic pain management (e.g., Catalano's workbook, Caudill's workbook, Aspen's pain management education manual, and Tollison's pain management patient guide)?
What are the costs (if available) associated with pharmacologic, technological and other interventions listed in questions 4, 5, and 6 above?
To our knowledge, there have been no systematic reviews specifically designed to answer any of the questions formulated by the CSCM.
Studying the natural history, prevalence, and treatment of neuropathic pain through the research evidence available to date was regarded as a complex task from the outset. The first major challenge for any effort to review the evidence from research on the management of pain following TSCI emanates from the nature of neuropathic pain, its definition, and our understanding of its underlying pathophysiology. Although there is a surprising degree of agreement around the definition of neuropathic pain within the clinical community, neuropathic pain is not a discrete physiologically measurable phenomenon, but a construct (or concept) derived from clinical observations (Portenoy, 1998). Although useful from the clinical perspective, the current construct of neuropathic pain is likely to be overly simplistic (Portenoy, 1998). In fact, there has been no confirmation that a constellation of "neuropathic mechanisms" actually exists (Portenoy, 1998). If such mechanisms exist, they are likely to be complex and highly variable, both within and across pain syndromes (Bennett, 1993; Portenoy, 1998). On the other hand, multiple mechanisms can be present within any given diagnostic category and even within individual patients (Fields and Rowbotham, 1994).
Another expected challenge to the application of the principles of evidence-based decision-making to the study of pain following TSCI is related to the amount and quality of the research data available. A recent analysis of specialized journals revealed that less than 1 percent of articles specifically addressed pain after TSCI (Siddall, Taylor, and Cousins, 1997). In addition, systematic reviews and empirical methodological studies have repeatedly shown that research in pain relief in general is incompletely reported, addresses few clinically relevant questions, and is prone to bias (Jadad and Cepeda, 1999). If these problems were also to be present in the small subset of research studies on pain after TSCI, the value of the little knowledge available would be reduced even further.
Lastly, other barriers were expected to emanate from the lack of a systematic approach to the management of neuropathic pain in general that could be applied to the management of pain following TSCI. To date, treatment decisions have been described as "largely hit or miss, mostly miss" (Fields, 1994), as clinicians have been forced to administer the same group of treatments to patients, regardless of the origin of the pain, hoping to find at least one that could give patients adequate relief (Max, 1990).
The team that developed the Task Order included a local executive team, a Technical Expert Panel (TEP), representatives from the CSCM (Appendix A) and the TOO. The TEP included a group of prominent professionals, purchasers, and patients who agreed to participate in this Task Order, as well as members of the CSCM.
By applying state-of-the-art methodology and by building on the accumulated experience of evidence-based decisionmaking in pain relief, this Task Order will not only support the production of guidelines by the CSCM and other interested organizations, but could also provide valuable information to support the decisions of clinicians, policy makers, researchers, advocates, consumers, journal editors, and any other group of individuals interested in the management of pain after TSCI.
In this chapter, we report our overall approach in conducting the systematic reviews included in this Task Order. The chapter starts with a brief description of the TEP, TOO, and partners at the CSCM and the process followed to interact with them throughout the development of the Task Order. The chapter concludes with details of the methodological approach that was followed to answer each of the questions formulated by the CSCM.
In subsequent chapters, we report the findings of systematic reviews of the available evidence relevant to each of such questions. When appropriate, we will provide additional information on specific methodological issues relevant to the question (s) addressed by each chapter.
The TEP for this Task Order included individuals who represented professional organizations, providers of health care, purchasers of health services, researchers, and consumers. The membership list is appended (Appendix B). These individuals are recognized as national and international leaders in the management of pain or in issues related to SCI. Given the number of questions, the wide variety of issues that they address, and the diversity of backgrounds among the large number of members of our TEP, we consulted with each of the panelists regarding their interest in focusing their attention on particular questions. We asked about their preferences, collated their responses, and ensured coverage of each question from at least two panelists, allocating them to areas of interest that matched their background. This communication strategy facilitated the topic refinement process and enabled us to approach all of the questions simultaneously and efficiently. We anticipated that this strategy also would assist in determining authorship of individual publications derived from this work. Members of our TEP were asked to nominate two or three peer reviewers.
We consulted our TOO, Dr. Harry Handelsman, and representatives of the CSCM at each step in the process. Mr. Paul Thomas and Ms. Dawn Sexton were the main contact people at the CSCM. Initially, the team of the MU-EPC communicated with the TOO and the representatives from the CSCM by telephone to discuss the goals and objectives of this Task Order. We also had a face-to-face meeting to complement the telephone discussions. The Chair of the Steering Committee of the CSCM, Dr. Kenneth Parsons, was consulted and agreed to become a member of the TEP. This facilitated collaboration with the CSCM and helped to ensure that we fully addressed the questions that formed the basis for this Evidence Report. Members of the CSCM were asked to identify potential peer reviewers for the draft Evidence Report.
A preliminary list of potential peer reviewers was drafted in October of 1999 and sent to the TOO. The final list of peer reviewers appears in Appendix C. In April of 2000, each proposed reviewer was contacted to request their assistance with the review of this draft Evidence Report and to provide a timeline and directions for their involvement. We delivered the report for their review in early July 2000. Their comments were returned to our criticisms editor, Dr. Pat Huston. The comments were then synthesized and returned to us for incorporation and editing of this final report.
Initially, we considered other groups of individuals with central neuropathic pain such as those with stroke or multiple sclerosis. However, in discussion with the TOO and the representatives from our partners, we established that the task order would focus upon adults and adolescents with CNP following TSCI.
The wording of some of the original questions asked for clinical recommendations, which are more appropriate for clinical practice guidelines than for an Evidence Report. For example, one question was originally stated as: "What are the best outcome measures and/or assessments that detect clinical changes in pain management for chronic central neuropathic pain?" Following discussions with our TOO and the representatives from our partners, the questions were reworded and reordered to ensure that they were clinically important and potentially answerable by research evidence. In answering the questions, the strength of evidence was summarized, but clinical recommendations were not made. The complete list of questions, which were grouped into those related to assessment, natural history, and interventions for treatment, is found in Chapter 1.
The last question formulated by the CSCM dealt with cost issues. Ideally, this question should have been answered in terms of the added value of quality of life relative to the costs of different management strategies. We searched our core electronic databases for studies that incorporated economic analyses or cost information. We also searched two economic databases, the National Health Services Economic Evaluations Database as well as the Health Economic Evaluations Database. We were unable to locate any studies specific to the management of CNP following SCI, therefore, this evidence report does not deal with cost issues. The TOO advised that performing an economic analysis was beyond the scope of the Task Order and, therefore, would not be conducted by us.
In consultation with the TEP and our partners, we developed a set of inclusion and exclusion criteria. Initially, we used very liberal selection criteria. We regarded as potentially eligible any article that described a study: (1) in humans; and (2) about the cause, management, or measurement of CNP. We excluded reports that were not primary studies; studies where the sample consisted of people without a TSCI, those without chronic neuropathic pain, or children younger than 13 years. We also excluded primary studies that did not contain data in the published report and studies in which the sample included individuals with TSCI as well as other types of CNP but where the results were not presented separately for individuals with TSCI.
Since there are nearly as many definitions of CNP as there are published reports, we did not attempt to provide a definition ourselves. We accepted any definition provided by the primary study authors. We also developed a tentative list of descriptions of CNP based on our preliminary searches. This list was used to guide the research team during the two-step screening process. We excluded studies that only used the term "chronic pain" without any other description of the pain experienced by the individuals in the study sample that could have helped us judge it as central and neuropathic.
Many studies included individuals with CNP and pathologies other than TSCI (e.g., multiple sclerosis, stroke). For the purposes of this Evidence Report and the evidence tables, we included only the results pertaining to those with TSCI. The only exception to this decision concerned adverse effects. If adverse effects were reported only for the entire sample, we indicated this in the evidence tables.
Studies of individuals with TSCI and CNP as well as reflex sympathetic dystrophy or post-traumatic syringomyelia were considered separately and are reported in the supplementary rather than in the primary evidence tables. We considered injuries to the cauda equina as peripheral nerve lesions rather than a central problem. It was uncertain if reflex sympathetic dystrophy, now referred to as complex regional pain syndrome, type 1 (CRPS-1) or complex regional pain syndrome, type 2 (CRPS-2), could be considered to be a form of CNP. Further, it is unclear if studies of individuals with pain associated with post-traumatic syringomyelia are sufficiently similar to those with CNP post TSCI who do not have a syrinx.
Treatment options were confined to conventional medicine. We did not seek out reports of nontraditional therapies. We anticipated that we would identify many studies of single or multiple case reports. While these studies can be important in determining adverse effects or generating hypotheses related to CNP, they are not useful to evaluate the effectiveness of interventions. For this reason, single or multiple case reports (usually with fewer than eight individuals with TSCI) are described in supplemental evidence tables within each chapter. Early in the screening process, we discovered that some larger studies of neuropathic pain included fewer than 10 people with TSCI. While results were reported separately for these individuals, the characteristics of the sample were provided for the entire group. Such studies also are reported in the supplemental evidence tables.
Citations of potentially eligible studies were identified through a systematic search of:
MEDLINE, EMBASE, PsycINFO were searched from the date of their release to end of May 2000.
CINAHL, HEALTHStar, Sociological Abstracts. These databases were searched from the date of their release to November 1999.
The reference lists of any eligible article identified in any of the above sources.
Personal files of all members of the local team and the TEP.
The development of the search strategy followed an iterative process. Initially, we chose search terms based on the MEDLINE indexing terms of several key publications. We then tested our initial search using the "See Related" function of PubMed to ensure that the search would retrieve the key publications previously identified. The search terms were then refined using the same process. The final MEDLINE search strategy appears in Appendix D. The MEDLINE strategy was modified to meet the specific features of CINAHL, EMBASE, and PsycINFO. We searched for unpublished studies by contacting members of the TEP. After discussion with the TOO, we decided not to attempt to contact study authors for additional information.
To identify relevant articles, we conducted a two-step screening process with six raters working in pairs. The first step was based upon the information available in titles and abstracts (where available) and was conducted by the same two raters working independently. Each citation was rated as "include," "exclude," or "unclear." A citation marked by either rater as "include" or "unclear" was retrieved. The terms sought during the first screening step were human, neuropathic pain, and spinal cord injuries. The second step of the screening process was based upon full text reports and involved all six raters randomly paired. Discrepancies were resolved by discussion. The form for the second screening is found in Appendix E.
Since we used different pairs of raters, we determined our agreement in applying the criteria for the second screening. The statistician on the McMaster Team, Dr. Charlie Goldsmith, designed a randomization schedule. Initially, all six raters screened the same set of 10 articles. The project assistant (who did not rate any articles), then assigned the articles sequentially following the unblinded randomization schedule to six raters for the first 90 articles. For the remaining articles, five raters were used.
In assessing the methodologic quality of the RCTs, we used a three-item validated scale, complemented with assessments of individual components supported by empirical methodological evidence (Jadad, Cook, Jones, et al., 1998). For observational studies evaluating effectiveness or safety of interventions, we used an instrument adapted from Downs and Black (Downs and Black, 1998). For studies on diagnosis or assessment of pain, we used the corresponding User's Guides to the Medical Literature (Jaeschke, Guyatt, and Sackett, 1994). We adapted the User's Guides on prognosis to evaluate studies on natural history (Laupacis, Wells, Richardson, et al., 1994). For studies that assessed predictors of outcome, we noted the comprehensiveness with which possible predictors were examined, the representativeness of the population studied, the extent to which assessments of predictor status and outcome were independent, and the completeness of followup. For the assessment of the quality of qualitative studies, we planned to use a tool adapted from Boulton et al. (Boulton, Fitzpatrick, and Swinburn, 1996).
All data extraction forms were developed, pilot-tested, and revised by members of the local research team including the team statisticians. After consultations with the TEP, our TOO, and our partners, the forms were approved for content. A general data extraction form was used with all studies, while individual forms were used for RCTs, observational studies, and case reports (Appendix E). Items related to study quality were embedded within the data extraction forms. Two reviewers completed data extraction independently for all studies except the case reports. For these studies, data were extracted by one reviewer and checked by another. Any disagreements were resolved by consensus. Following consensus on each item, the data forms were scanned into a Microsoft Access database using Teleform software.1
Descriptive statistics were calculated for all fields of the database. Evidence tables were constructed to describe the most salient features of the included studies according to the review question. These tables are found at the end of each chapter along with the relevant supplementary tables. The tables have the following general structure:
a summary of the key characteristics of the studies reviewed, including the name of the first author, the year of publication of the study, the country where the study was centered, the type of study design, and items related to quality;
items related to the characteristics of the sample, including the number of patients allocated, level and completeness of the injury, time since injury, onset of pain, duration of pain, and description of pain;
the interventions studied and the duration of exposure to each of the interventions;
the outcomes of interest measured (following the list provided above);
the key results for the outcomes of interest; and
adverse effects.
To make the data extracted easier to understand, the information on the studies was organized in the tables following alphabetical order. The studies are also presented in alphabetical order within the text. In addition, every effort was made to describe the studies according to their methodological quality, taking into account both the evidence-based criteria and the elements that were not supported by empirical evidence.
The local research team at the MU-EPC, in consultation with members of the partner organization and the TOO, evaluated the overall quantity and quality of the data available. This evaluation led to the conclusion that meta-analysis would be inappropriate to summarize the evidence on each of the research questions or for each of the main categories of interest. The main reasons for this decision were substantial clinical heterogeneity across the studies (e.g., interventions evaluated, patient samples, duration); inconsistency in outcome measurements; low methodological quality; and incomplete data reporting (see detailed descriptions within each category). The use of meta-analysis to synthesize this type of data has been associated with a greater chance of obtaining imprecise and potentially misleading results (Ioannidis, Cappelleri, and Lau, 1998). Therefore, this report represents a systematic qualitative review of the existing evidence, emphasizing the implications for clinical practice and the directions that future researchers could take to fill existing knowledge gaps.
Every effort was made to present the information obtained in each of the categories following a uniform format. Overall, each section begins with a general description of the most salient characteristics of the studies and ends with a summary of the main findings of the studies. The way in which the results from the individual studies are presented varies substantially across chapters. This reflects the different perspectives that the questions provide on the management of CNP following TSCI.
After considering multiple reports and several levels of screening, there were 132 unique studies forming the basis for the Evidence Report. Of these, there were six randomized trials and 126 observational studies, including 47 case series and at least 56 single or multiple case reports.
Single or multiple case reports (usually with fewer than eight individuals with TSCI) are described in supplemental evidence tables within each chapter and are not included in the characteristics featured below. The most salient characteristics of all the studies included (Evidence Tables 3.6.5 to 3.6.29) were the following:
The number of studies has been doubling in less than 10 years since 1965.
Sixty-one percent of the reports did not report any information about the source of funding for the research. Government supplied funding in 22 percent of the reports, and charities provided funding for 9 percent.
Most papers (68) were published in the United States, while the United Kingdom published 12. No other country had greater than 7 (Germany had 7) studies included.
Case reports (56) and case series (47) accounted for 78 percent of all included reports. There were 10 surveys included and 6 RCTS, 6 validation, 4 nonrandomized controlled trials, and 3 case-control studies.
Excluding the case reports, across the remaining 75 studies, there were 3,873 patients. In one study, the number of included patients was not clear.
There were over 100 patients included in the study sample of 17 studies; 30 percent of the studies had 25 or fewer patients.
In four studies, there were more than 100 patients with TSCI/CNP included; 35 studies had 20 or fewer patients with TSCI/CNP.
Studies conducted in a tertiary care setting accounted for 82 percent of the reports.
Eligibility criteria were explicitly reported in 34 percent of the papers.
Preplanning of the study duration was clearly reported in 49 percent of the papers, with 18 studies using a time cutoff and 29 studies using another endpoint. Times used varied from 1 week to 50 months.
Seventy-four percent of the studies included just one treatment group.
Forty-two percent of the studies included patients with cervical injury, and 49 percent included patients with thoracic injury. In 58 percent of studies, the completeness of the injury was not reported; and in 33 percent of the studies, there were multiple levels of injury reported.
Five studies used the American Spinal Injuries Association (ASIA) classification scale, and six studies used Frankel's Classification scale.
The most common causes of injury were road traffic accidents, falls, and gunshot wounds.
Seventy percent of the studies did not report whether surgical stabilization was used, and 29 percent reported some surgical stabilization.
Fifty-four percent of the reports did not specify the length of time since injury.
Forty-one percent of the reports did not specify the age of participants. Of those that did specify, 45 percent had participants in the age group between 30 and 50 years.
Sixty-two percent of the authors did not define neuropathic pain.
The most common descriptor for pain was "burning," with 53 percent of studies reporting usage of the term. The second most common was "dysesthesia," reported in 24 percent of studies.
The length of time between injury and start of pain was not reported in 83 percent of studies. The most frequently cited length of time was between 1 month and 6 months (12%).
The duration of pain was not reported in 82 percent of studies, and the area of the body affected by the pain was not reported in 57 percent of studies.
The most commonly used outcome measures were: interview or narrative about the pain (29%), visual analogue scale (26%), and the McGill Pain Questionnaire (16%).
In keeping with the questions proposed by our partners, the studies featured in subsequent chapters of this Evidence Report include:
Assessment and natural history: 58 unique studies, of which 20 were case reports or case series with fewer than 8 patients.
Pharmacological interventions of interest: 27 unique studies, of which 15 were case reports or case series with fewer than 8 patients.
Spinal cord and deep brain stimulation: 25 unique studies, of which 17 were case reports or case series with fewer than 8 patients.
DREZ lesions and other surgical interventions: 23 unique studies, of which 8 were case reports.
There were no studies evaluating the role of pharmacological algorithms or multidisciplinary approaches, and two studies assessed a self-management program in patients with CNP following TSCI. One study focused on relaxation techniques in four single case reports (Grzesiak, 1977) and the second on laughter and was a pilot study of 11 patients (Henderson and Mowry, 1995) (see Evidence Tables 3.1.30 and 3.1.31). No adverse effects or safety issues were found in either study. No other studies were located that addressed self-management approaches.
Three of the most complex and overlooked aspects of pain after TSCI are the diagnosis, assessment, and study of the natural history of CNP. To a large extent, the complexity associated with these aspects of CNP stems from the nature of the pain itself and from the inherent difficulty in defining and classifying it.
As with other types of pain, CNP remains a construct or concept, derived from clinical observations and patients' reports, that cannot be independently and objectively verified or measured by laboratory tests. Even if clinicians agree that the presentation of a pain is "neuropathic," there can be no independent verification of a neuropathic pathophysiology in the clinical setting. To date, there has been no confirmation that a constellation of "neuropathic mechanisms" exists (Portenoy, 1998). If such mechanisms exist, they are likely to be complex and highly variable, both within and across pain syndromes (Bennett, 1993; Portenoy, 1998). On the other hand, multiple mechanisms can be present within any given diagnostic category and even within individual patients (Fields and Rowbotham, 1994).
The corollary is that, although useful from the clinical perspective, the current construct of CNP is overly simplistic (Portenoy, 1998) and has limited value during the development of valid and accurate diagnostic criteria. As a result, most of the systems available to classify CNP typically rely on clinical interpretation of pain location (e.g., pain at or below the level of the neurological lesion), the anatomic basis of pain (e.g., visceral or root pain), or its putative pathophysiology (e.g., mechanical or psychological) (Burke, 1973; Donovan, Dimitrijevic, Dahm, et al., 1982; Siddall, Taylor, and Cousins, 1997). These systems retain the advantage of clinical simplicity and have been used to guide therapy. These pain classifications, however, require the simultaneous consideration of several different pain constructs of which there is little understanding. Another limitation of the classification schemes used to diagnose CNP (discriminative tools) is that they are not useful for measurement of pain severity or functional impact, which require evaluative instruments. There are numerous pain measurement instruments that have been developed for the quantification of pain, usually for the purposes of performing therapeutic studies (Jadad, Moore, Carroll, et al., 1996). Several of these instruments are generic tools that have been employed in the specific setting of TSCI. Others are likely to be developed exclusively for use in studies including people with pain following TSCI. Regardless of the initial purpose behind their development, the use of these instruments for discriminative or evaluative purposes ideally requires the establishment of acceptable psychometric properties (validity, reliability, and responsiveness) in the specific context of interest. For example, an instrument that is valid and reliable for measurement of neuropathic pain secondary to diabetic peripheral neuropathy may not be valid and reliable for individuals with CNP after TSCI.
For these reasons, the CSCM included questions pertaining to discriminative and evaluative approaches and the natural history of CNP as part of the review of the management of CNP after TSCI. To our knowledge, there have been no systematic efforts to identify, collect, synthesize, and distill the research evidence on the assessment, diagnosis, and natural history of CNP after TSCI. In this chapter, we describe the results of such an effort.
A complete description of the general methods of the review is described in Chapter 2. In the next section, we describe our approach to addressing the questions related to assessment or natural history of CNP in TSCI.
This chapter includes the evaluation of the literature to address the first two groups of questions proposed by the CSCM regarding the assessment and natural history of CNP in TSCI. These questions were:
What are the measurement properties of:
assessment approaches for chronic CNP per se (including criteria and tools such as inventories, questionnaires, and scales);
other outcome measures or assessments (related to the experience of pain); and
assessment approaches (including criteria and tools such as inventories, questionnaires, and scales) to identify new onset musculoskeletal pain against a background of chronic CNP?
What is the strength of evidence for strategies for the differential diagnosis of chronic CNP from other types of pain?
What is the strength of evidence for identifying the prevalence of acute and chronic neuropathic pain and factors that could predict the development of chronic CNP?
In addition to the generic methods described in Chapter 2, two reviewers (DW, MAO) assessed each study (using the form found in Appendix E) and categorized the studies into each of the following groups.
The literature search yielded 158 publications that addressed chronic CNP in TSCI. There were 65 citations that provided 58 unique reports addressing assessment and natural history questions. Of these, 7/58 (14%) addressed diagnosis or assessment; 2/58 (3%) addressed validation of an outcome tool or measure for CNP; 35/58 (60%) addressed factors that are associated with, predict, or addressed the prevalence of CNP; and 17/58 (29%) generated or tested hypotheses related to CNP. Some reports addressed more than one question.
There were methodological limitations present in all five of the studies. Neuropathic pain was partially or fully defined in three of the publications (Cohen, McArthur, Vulpe, et al., 1988; Davidoff, Roth, Guarracini, et al., 1987; Frisbie and Aguilera, 1990). The sampling method was either not reported or a convenience sample was used in each study, limiting the ability to assess the representativeness of each sample. Although three of the assessment studies used more than one instrument to measure pain severity (Beric, Dimitrijevic, and Light, 1992; Davidoff, Roth, Guarracini, et al., 1987; Quigley and Veit, 1996), none designated one instrument to be a reference standard in an attempt to determine construct validity.
One study was designed to determine how reliably the clinical history could discriminate CNP from other pain syndromes (Frisbie and Aguilera, 1990); and the others were designed to prospectively assess the quality or intensity of CNP in individuals with TSCI (Beric, Dimitrijevic, and Light, 1992; Cohen, McArthur, Vulpe, et al., 1988; Davidoff, Roth, Guarracini, et al., 1987; Quigley and Veit, 1996).
In 1990, Frisbie et al. (Frisbie and Aguilera, 1990) determined that two clinical history points -- pain quality and pain location relative to the level of paralysis -- were 100 percent reliable (sensitive) for CNP diagnosis (rather than musculoskeletal or syringomyelia pain) when judged by the standard of the absence of structural pathology found by clinical examination or radiography. Neuropathic pain was defined in this study as "neurogenic," based on adjectival description and being located at or distal to the level of paralysis. This study was not blinded and all assessments were done by the same clinicians, potentially increasing the possibility of bias.
Three publications (Cohen, McArthur, Vulpe, et al., 1988; Davidoff, Roth, Guarracini, et al., 1987; Quigley and Veit, 1996) used all or part of the McGill Pain Questionnaire (MPQ) to assess the presence and severity of CNP. Other instruments used included visual analog scales, the Sternbach Pain Index, the Zung Pain and Distress Scale (PAD), and pain drawings. All of the studies but one (Davidoff, Roth, Guarracini, et al., 1987) were prospective. In this study, patients with the symptoms of chronic CNP were compared to patients in a database. Cohen et al., 1988 compared patients with chronic CNP following a TSCI to those without a spinal cord injury. In some studies, pain was judged as being more severe in quadriplegic individuals but more common in those with paraplegia (Davidoff, Roth, Guarracini, et al., 1987; Quigley and Veit, 1996). Pain syndromes resulting from both complete and incomplete TSCI were generally more severe than chronic pain syndromes from other causes (Davidoff, Roth, Guarracini, et al., 1987). One study provided data on the correlations between different outcome measures: the Number of Words Chosen subscale of the MPQ correlated positively with the results of the PAD (Davidoff, Roth, Guarracini, et al., 1987).
Patients who were older at injury or at the time of study had a higher prevalence of CNP (Anke, Stenehjem, and Stanghelle, 1995; Demirel, Yllmaz, Gencosmanoglu, et al., 1998; Jung, 1990; Levi, Hultling, Nash, et al., 1995; Nepomuceno, Fine, Richards, et al., 1979). The presence of CNP was associated with incomplete spinal cord injuries in three publications (Demirel, Yllmaz, Gencosmanoglu, et al., 1998; Jung, 1990; Siddall, Taylor, McClelland, et al., 1999) but not in three others (Anke, Stenehjem, and Stanghelle, 1995; Levi, Hultling, Nash, et al., 1995; Summers, Rapoff, Varghese, et al., 1991). The level of spinal cord injury (thoracic level or paraplegia rather than quadriplegia) was associated with higher pain prevalence in some studies (Davidoff, Roth, Guarracini, et al., 1987; Demirel, Yllmaz, Gencosmanoglu, et al., 1998; Fenollosa, Pallares, Cervera, et al., 1993; Kennedy, Frankel, Gardner, et al., 1997; Rintala, Loubser, Castro, et al., 1998; Siddall, Taylor, McClelland, et al., 1999) but not others (Anke, Stenehjem, and Stanghelle, 1995; Levi, Hultling, Nash, et al., 1995). The presence of chronic pain was not associated with gender (Anke, Stenehjem, and Stanghelle, 1995; Fenollosa, Pallares, Cervera, et al., 1993; Levi, Hultling, Nash, et al., 1995), time since injury (Jung, 1990; Levi, Hultling, Nash, et al., 1995), or pain at 6 weeks postinjury (Demirel, Yllmaz, Gencosmanoglu, et al., 1998). Gunshot wounds were associated with a higher prevalence of CNP (Richards, Stover, and Jaworski, 1990; Rintala, Loubser, Castro, et al., 1998). Surgical intervention was not associated with CNP in two studies (Siddall, Taylor, McClelland, et al., 1999; Summers, Rapoff, Varghese, et al., 1991) but in another, patients with CNP were less likely to have received surgery (Davidoff, Roth, Guarracini, et al., 1987).
The presence of CNP may be associated with certain pathological changes in the spinal cord, such as extended atrophy, malacia, and syrinx formation, that are detectable by magnetic resonance imaging (Wang, Bodley, Sett, et al., 1996).
Higher pain severity was associated with incomplete lesions in one study (Rintala, Loubser, Castro, et al., 1998) but not in another (Summers, Rapoff, Varghese, et al., 1991). Pain severity was not related to the level of injury or surgical intervention, except in one study where quadriplegics had more intense pain than paraplegics (Summers, Rapoff, Varghese, et al., 1991). Gunshot wounds and other violent etiologies were associated with increased pain levels (Richards, Stover, and Jaworski, 1990; Rintala, Loubser, Castro, et al., 1998), but etiology was not an important factor in other studies (Cairns, Adkins, and Scott, 1996; Fenollosa, Pallares, Cervera, et al., 1993). There was no relationship between pain severity scores and the time from injury (Jung, 1990) or the duration of pain symptoms (Davidoff, Roth, Guarracini, et al., 1987).
Several studies assessed the relationship of psychological and social factors with chronic pain. The presence of pain was associated with worse quality of life (Anke, Stenehjem, and Stanghelle, 1995; Stensman, 1994; Svendsen, Drewes, Biering-Sorensen, et al., 1993), increased depression and stress (Kennedy, Frankel, Gardner, et al., 1997), and worse self-assessed health (Cairns, Adkins, and Scott, 1996; Rintala, Loubser, Castro, et al., 1998). Increased pain severity was associated with higher levels of anxiety, stress, depression, and anger (Demirel, Yllmaz, Gencosmanoglu, et al., 1998; Nepomuceno, Fine, Richards, et al., 1979; Rintala, Loubser, Castro, et al., 1998; Stormer, Gerner, Gruninger, et al., 1997; Summers, Rapoff, Varghese, et al., 1991). Coping strategies were found to be important predictors of pain after SCI (Jung, 1990; Stensman, 1994; Stormer, Gerner, Gruninger, et al., 1997).
The failure to use a clear definition of CNP, use of unvalidated outcome instruments, and inadequate sampling techniques threaten the validity of most of these association studies and largely prevents between-study comparisons of study results.
Seventeen studies contained a complete or partial definition of neuropathic pain or referenced another source. Seventeen studies were prospective. The cross-sectional methods employed included surveys done either in the clinic or at home (Cairns, Adkins, and Scott, 1996; Fenollosa, Pallares, Cervera, et al., 1993; Jung, 1990; Nepomuceno, Fine, Richards, et al., 1979; Rintala, Loubser, Castro, et al., 1998; Rose, Robinson, Ells, et al., 1988; Svendsen, Drewes, Biering-Sorensen, et al., 1993; Turner and Cardenas, 1999; Waisbrod, Hansen, and Gerbershagen, 1984) or personal or telephone interview (Anke, Stenehjem, and Stanghelle, 1995; Beric, 1990; Davidoff, Guarracini, Roth, et al., 1987; Kennedy, Frankel, Gardner, et al., 1997; Levi, Hultling, Nash, et al., 1995; New, Lim, Hill, et al., 1997; Richards, Stover, and Jaworski, 1990; Siddall, Taylor, McClelland, et al., 1999; Stensman, 1994; Stormer, Gerner, Gruninger, et al., 1997; Summers, Rapoff, Varghese, et al., 1991; Wang, Bodley, Sett, et al., 1996; Woolsey, 1986).
Five studies evaluated at least some patients in the sample on more than one occasion (Burke, 1973; Cairns, Adkins, and Scott, 1996; New, Lim, Hill, et al., 1997; Siddall, Taylor, McClelland, et al., 1999; Stensman, 1994). The response rates ranged from 37 to 93 percent for surveys and from 39 to 100 percent for interviews. The sampling frame was defined clearly in seven studies (Fenollosa, Pallares, Cervera, et al., 1993; Jung, 1990; Levi, Hultling, and Seiger, 1995; Rintala, Loubser, Castro, et al., 1998; Rose, Robinson, Ells, et al., 1988; Turner and Cardenas, 1999; Wang, Bodley, Sett, et al., 1996); most other studies used samples of convenience from rehabilitation hospitals or surveyed from populations in which the denominator was not estimable.
Acute pain (within the first 8 weeks postinjury) was specifically measured in all individuals in two studies (Kennedy, Frankel, Gardner, et al., 1997; Sved, Siddall, McClelland, et al., 1997); the individuals in these studies were also evaluated later to determine the prevalence of CNP. Five other studies included some individuals with acute pain due to the range of postinjury assessments included (Anke, Stenehjem, and Stanghelle, 1995; Cairns, Adkins, and Scott, 1996; Demirel, Yllmaz, Gencosmanoglu, et al., 1998; Long, 1976; New, Lim, Hill, et al., 1997). The remaining studies included only patients evaluated more than 8 weeks after TSCI.
The prevalence of acute or early neuropathic pain (within the first 8 weeks) depended on the time of assessment and definition of CNP. At 2 weeks in one study, 38 percent of individuals had pain at the level of the lesion and 14 percent had pain below the lesion (Sved, Siddall, McClelland, et al., 1997). In another study, 80 percent experienced pain at 6 weeks post-injury, but CNP was not defined (Kennedy, Frankel, Gardner, et al., 1997). The remaining studies that included some individuals during the acute postinjury stage estimated the prevalence of pain at admission to a rehabilitation facility to be 8 to 76 percent.
Two studies evaluated neuropathic pain prevalence longitudinally and extended to at least 1 year postinjury (Kennedy, Frankel, Gardner, et al., 1997; Siddall, Taylor, McClelland, et al., 1999). The prevalence of neuropathic pain at the level of the neurological lesion was relatively stable during the first postinjury year (38% at 2 weeks, 38% at 26 weeks, and 28% at 52 weeks), but below-lesion pain increased sharply between 26 weeks (19%) and 52 weeks (50%) (Siddall, Taylor, McClelland, et al., 1999). Another study, however, showed that "pain" prevalence was 80 percent at 6 weeks and fell to 63 percent at 1 year; neuropathic pain was not defined in this study (Kennedy, Frankel, Gardner, et al., 1997).
In sum, these publications demonstrate that neuropathic pain begins shortly after TSCI and is a common problem even many years after the injury. There are several methodological problems with this group of articles that resulted in the wide range of prevalence estimates. In many cases, the description of the sampling frame was unclear and selection bias was likely. In other studies, it was uncertain what the appropriate denominator was for prevalence calculation. The lack of uniform neuropathic pain definition, low response rates, inadequate sampling techniques, and variable time from injury until pain assessment all contributed to heterogeneous prevalence assessments.
The findings of this systematic review demonstrate that many investigators have addressed important assessment and natural history issues regarding CNP in TSCI. Neuropathic pain is common in individuals who suffer TSCI, both in the acute (first 8 weeks postinjury) and chronic phases, with most prevalence estimates ranging from 40 percent to 75 percent for chronic pain. This pain is moderate to severe in 25 percent to 60 percent of these individuals and may interfere with activities of daily living. An increased likelihood of developing CNP may be associated with certain factors such as older age and having a gunshot wound as the cause of SCI. The presence and severity of CNP appears to be associated with anxiety, depression, and lower perceived health status. The description and location of pain may allow accurate diagnosis of CNP. There are no discriminative or evaluative measurement instruments that have been adequately investigated with respect to psychometric measurement properties in this setting.
We noted a wide range of definitions of CNP in samples of patients with TSCI. Some authors explicitly defined CNP using their own definition, others referred to previously described classification systems, and some simply used the labels "spinal cord injury pain" or "spinal cord pain." This disparity results in uncertainty regarding what type of pain syndrome a given sample has, complicates interpretation of results, and impairs the ability to replicate the study.
There were very few studies that systematically evaluated the psychometric properties (validity, reliability, responsiveness) of a discriminative or evaluative instrument for the measurement of CNP in individuals with TSCI. Several authors utilized pain measurement instruments that have been validated in other settings but provided predominantly descriptive information concerning the subjects they studied. These instruments, however, may have different properties when used to assess CNP after TSCI. There were no studies that defined a CNP construct, identified a reference standard instrument, or attempted to assess the validity of the construct.
The studies of prevalence of CNP in TSCI patients varied strikingly in methodology. Most studies did not adequately define CNP, describe the sampling frame in sufficient detail, or provide an estimate of an appropriate denominator. Many of the studies used a sample of convenience or consecutive patients admitted to one rehabilitation center, possibly resulting in selection bias. These reasons, as well as the use of different outcome measures and evaluations at different stages of rehabilitation, led to the wide range of prevalence estimates.
The results of this systematic review support the notion that CNP is very prevalent in individuals after TSCI and warrants specific inquiry by health care professionals caring for such individuals. It is associated with several psychological and psychiatric comorbidities and is often severe enough to impair or prevent optimal physical function and daily activities.
The results of this systematic review demonstrate that improvements are needed in defining the construct of chronic neuropathic pain in the setting of SCI. The most commonly used classifications of CNP (Donovan, Dimitrijevic, Dahm, et al., 1982; Siddall, Taylor, and Cousins, 1997) and the ad hoc definitions used in many other studies emphasize interpretation of the source of the pain (neurogenic versus visceral, root, musculoskeletal, or otherwise) and its location (at or below the level of neurological injury). The construct of neuropathic pain used in these classification systems is therefore multidimensional and subject to individual interpretation. The strength of current concepts of CNP definitions is their widespread acceptance as constructs. The number of different classification systems and their heterogeneity represents a weakness of currently available research.
The measurement properties of instruments used to evaluate neuropathic pain have not been defined. Well-designed experimental and observational studies are needed to establish construct validity for neuropathic pain definitions and for instruments that assess pain severity. This improvement would allow for better prevalence and association studies. Without this advancement, studies that generate or test new hypotheses concerning the pathophysiology of neuropathic pain will not be replicable. Test-retest reliability and responsiveness must also be established in order for future research to advance beyond the descriptive stage towards valid evaluative and therapeutic studies.
Future studies of assessment and diagnostic tools may benefit from improvements in study design. Definition of the sampling frame and methods of sampling that reduce selection bias would strengthen prevalence and association studies. Blinded outcome assessment, which was utilized by a small number of studies we identified, would also strengthen validity studies.
Pharmacological interventions are usually the first line during the treatment of patients with CNP of any origin. The number of pharmacological interventions with proven effectiveness, however, is so small that clinicians and patients are usually forced to adopt a "trial and error" policy, running through the same categories of treatments regardless of the origin of the pain (Max, 1990). These categories often include NSAIDs, opioids, antidepressants, anticonvulsants, alpha-2-agonists, steroids, and local anesthetics. Clinicians and patients not only have a wide variety of choices within each of these categories, but also can expand the number of options through the use of drug combinations, different routes of administration, novel formulations, and sometimes very sophisticated delivery systems. Progress in our understanding of the mechanisms responsible for neuropathic pain has led to the development of new compounds, most of which remain under development and evaluation.
To our understanding, there are no systematic reviews designed specifically to evaluate the effectiveness and safety of established or emerging interventions used for the treatment of CNP following TSCI.
The objective of this chapter is to describe the results of a systematic review of studies addressing the following questions, all of which were listed under Group III in the questions proposed by the CSCM:
What is the evidence for the effectiveness and safety of each of the following classes of medications: simple analgesics (including NSAIDs and acetaminophen); antidepressants (including tricyclics and SSRIs); antiseizure medication; narcotics; muscle relaxants; NMDA antagonists; and local anesthetics?
How do these classes of medications compare with each other?
What is the strength of evidence for the effectiveness and safety of treatment algorithms including these classes of medication?
No studies were found that addressed the last questions. A complete description of the general methods of the review is found in Chapter 2.
A total of 31 potentially eligible studies were identified. Three studies were excluded during the data extraction phase (Attal, Gaudé, Brasseur, et al., 2000) because the study was not about the management of neuropathic pain (Hansebout, Blight, Fawcett, et al., 1993) or because data pertaining to patients with TSCI could not be extracted (Attal, Gaudé, Brasseur, et al., 2000; Bravo, Labarta, and Garcia, 1988). One study (Taira, Kawamura, Tanikawa, et al., 1995) was published twice in two different journals.
A total of 12 independent studies were included in the review (Chiou-Tan, Tuel, Johnson, et al., 1996; Davidoff, Guarracini, Roth, et al., 1987; Drewes, Andreasen, and Poulsen, 1994; Eide, Stubhaug, and Stenehjem, 1995; Epstein and Childers, 1998; Erzurumlu, Dursun, Gunduz, et al., 1996; Fenollosa, Pallares, Cervera, et al., 1993; Glynn, Jamous, and Teddy, 1992; Glynn, Jamous, Teddy, et al., 1986; Heilporn, 1978; Loubser and Akman, 1996; Loubser and Donovan, 1991). Of these, five were RCTs (Chiou-Tan, Tuel, Johnson, et al., 1996; Davidoff, Roth, Guarracini, et al., 1987; Drewes, Andreasen, and Poulsen, 1994; Eide, Stubhaug, and Stenehjem, 1995; Loubser and Donovan, 1991), two were non-RCTs with contemporaneous controls (Erzurumlu, Dursun, Gunduz, et al., 1996; Glynn, Jamous, Teddy, et al., 1986), and seven were case series with more than eight patients (Davidoff, Guarracini, Roth, et al., 1987; Epstein and Childers, 1998; Fenollosa, Pallares, Cervera, et al., 1993; Glynn, Jamous, and Teddy, 1992; Heilporn, 1978; Loubser, 1997; Loubser and Akman, 1996).
All five RCTs included a placebo group. All but one of the RCTs included only one active group. This RCT compared ketamine, alfentanyl, and placebo (Eide, Stubhaug, and Stenehjem, 1995).
There were three studies evaluating opioids. One of them was an RCT (Eide, Stubhaug, and Stenehjem, 1995), one a nonrandomized comparative single-blind crossover study (Glynn, Jamous, Teddy, et al., 1986), and one was a case series (Fenollosa, Pallares, Cervera, et al., 1993).
Two studies evaluated anticonvulsants. One of them was an RCT (Drewes, Andreasen, and Poulsen, 1994) and the other a case series (Epstein and Childers, 1998).
Two studies evaluated local anesthetics. Both of these studies were RCTs with crossover designs (Chiou-Tan, Tuel, Johnson, et al., 1996; Loubser and Donovan, 1991).
Two studies performed by the same research group evaluated the effect of clonidine in patients with pain after TSCI. One of the studies was a nonrandomized comparative single-blind crossover study (Glynn, Jamous, Teddy, et al., 1986), and the other was a case series (Glynn, Jamous, and Teddy, 1992).
One study each evaluated the effect of baclofen, trazodone, and ketamine for pain after TSCI.
There were three studies that included drug combinations. One of these studies was a non-RCT comparative study with contemporaneous controls (Erzurumlu, Dursun, Gunduz, et al., 1996). The other two were case series (Fenollosa, Pallares, Cervera, et al., 1993; Heilporn, 1978).
It was disappointing to find so few studies evaluating the effect of pharmacological interventions for CNP after TSCI. The small sample sizes, the poor methodological quality, and the incomplete reporting of the studies available compounded this dearth of research efforts.
The main conclusion from the limited data identified in this review is that it is difficult to determine whether pharmacological interventions have a role in the management of CNP after TSCI. There was only one small study on antidepressants, which are regarded as one of the cornerstones of the management of other types of neuropathic pain. This study evaluated the effect of trazodone and did not show a statistically significant difference with placebo. The lack of more studies on antidepressants as a single therapy, particularly some evaluating tricyclic compounds, limits any attempt to judge the value of this group of interventions in patients with CNP after TSCI.
A similar situation was found in relation to anticonvulsants. Two studies, one of which was a case series, had small sample sizes and such incomplete reports that it was impossible to determine the value of this group of interventions, which is also regarded as very important for the treatment of other types of neuropathic pain.
There was some evidence of effectiveness for local anesthetics given intrathecally. The only study available showed that lidocaine led to a statistically significant difference in the number of patients who reported reduction in pain intensity compared with placebo. The sample size of this study was small and the double-blinding questionable. Mexiletine, a local anesthetic-like drug, showed no benefit compared with placebo when given orally to 11 patients.
The situation was similarly disappointing in relation to alpha-2-adrenergic agonists. We found two studies, with 25 patients in total, showing some promise for clonidine. No firm conclusion, however, can be derived from these small studies.
Opioids were evaluated in three studies with a total of 36 patients. The drugs and routes of administration used in the studies were variable. The only RCT showed reduction in pain with alfentanyl in nine patients. The other studies also showed response of pain after the administration of morphine epidurally and intrathecally.
Studies of drug combinations were either poorly reported or designed to answer questions of very limited clinical relevance.
In sum, research on pharmacological interventions for the management of CNP after TSCI is in its infancy. The evidence available is so limited that it is impossible to draw any conclusions regarding their role in clinical practice. Although it appears that local anesthetics, opioids, and clonidine given spinally may play a role in the management of CNP after TSCI, the evidence available comes from few, small, poorly reported, and largely uncontrolled studies.
While the needed evidence is gathered directly from patients with CNP after TSCI, clinicians interested in using pharmacological interventions will have to rely on research on these interventions in other patient populations.
Several invasive approaches have been used to treat refractory chronic pain conditions such as CNP following TSCI. Initially, these included destructive neurosurgical techniques such as posterior and anterior rhizotomies, neurectomies, radiofrequency thermocoagulation, or cryosurgical destruction of spinal nerves (Richardson, 1987). With increasing understanding of the complexity and plasticity of the spinal cord physiology, it was realized that it might be possible to obtain clinically useful analgesia with techniques that did not require destruction of neural tissue. These techniques involve stimulation of the spinal cord (spinal cord stimulation [SCS]) or specific sites within the brain (deep brain stimulation [DBS]). Both therapies are usually reserved for selected patients with chronic intractable pain shown to be resistant to conventional noninvasive therapies. Both involve the placement of electrodes over nervous tissue and electrical stimulation of underlying structures to produce pain relief. Both techniques also have the advantage of allowing assessment of responsiveness to therapy prior to permanent implantation (Nguyen, Keravel, Feve, et al., 1997; Richardson, Meyer, and Cerullo, 1980). DBS is generally used as a last resort for selected patients with severe pain after failure of SCS or in cases where SCS is contraindicated (Young and Rinaldi, 1994).
The precise mechanisms by which pain relief may be achieved using SCS or DBS are unclear. The theory underlying SCS involves the selective stimulation of large Aα and Aβ fibers to inhibit the nociceptive input from smaller Aδ and C fibers closing the spinal "gate." Stimulation of dorsal nerve roots and the dorsal columns is thought to activate inhibitory networks within the spinal cord that prevent transmission of pain (Meglio, Cioni, and Rossi, 1989). SCS may also result in the activation of descending inhibitory pathways as well as the release of neurochemical mediators that suppress pain (Richardson, 1987). In turn, stimulation of specific deep brain structures is thought to produce inhibition of nociceptive pathways within the brain as well as modulation of descending inhibitory pathways (Levy, Lamb, and Adams, 1987; Young and Rinaldi, 1994).
This chapter also includes a brief discussion of two studies of transcutaneous nerve stimulation (TNS) (Davis and Lentini, 1974; Doctor, 1996).
Spinal cord stimulation may be achieved via percutaneous insertion of electrodes into the epidural space near new roots corresponding to painful dermatomes. Potential responsiveness to therapy is generally associated with superimposition of stimulation-related paresthesias or pain relief over pain segments on testing. If responsiveness is shown, a receiver is implanted that allows the patient to regulate pain relief. Surgical placement of electrodes by laminectomy is generally reserved for patients in whom percutaneous placement is not possible.
Deep brain stimulation involves surgical placement of electrodes over selected sites within the brain. Deafferentation pain has been treated by stimulation of several sites within the thalamus, the internal capsule, and the motor cortex (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Lefaucheur, Decq, et al., 1999). The electrodes may be placed using several techniques, including positioning them into the intracranial epidural space via a burr hole or by craniotomy (Nguyen, Lefaucheur, Decq, et al., 1999). Pain relief after electrode placement is regulated by the patient in a manner similar to SCS.
Transcutaneous nerve stimulation involves the placement of electrodes placed over the skin to deliver electrical bursts to cutaneous and deep mechanoreceptors and is thought to affect the perception of pain. Treatment parameters can be altered by varying the frequency, amplitude, and voltage. The effectiveness of TNS for the treatment of chronic pain has been debated, and its effect on CNP following TSCI is unknown (Gadsby and Flowerdew, 2000).
The specific search strategies and methods used for data extraction are summarized in Chapter 2.
A total of 33 eligible reports were identified and retrieved. Seventeen unique studies reported on the effectiveness of SCS (Broggi, Servello, Dones, et al., 1994; Buchhaas, Koulousakis, and Nittner, 1989; Cole, Illis, and Sedgwick, 1991; Devulder, Vermeulen, De Colvenaer, et al., 1991; Kumar, Nath, and Wyant, 1991; Lazorthes, Verdie, and Arbus, 1978; Long and Erickson, 1975; Meglio, Cioni, Prezioso, et al., 1989; Moraci, Ambrosio, and Mignini, 1982; Nielson, Adams, and Hosobuchi, 1975; North, Ewend, Lawton, et al., 1991; Richardson, 1987; Richardson, Meyer, and Cerullo, 1980; Shimoji, Hokari, Kano, et al., 1993; Simpson, 1991; Tsuda and Tasker, 1985; Urban and Nashold Jr., 1978), six evaluated DBS (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999; Young, Tronnier, and Rinaldi, 1992), and two evaluated TNS (Davis and Lentini, 1974; Doctor, 1996).
The general quality of studies was poor. The reports consisted of noncomparative observational studies or case reports and contained information specific to patients with TSCI that was often inextricably intermeshed with that of other patients treated for chronic pain of different etiologies so that data extraction was impossible (Broggi, Servello, Dones, et al., 1994; Nielson, Adams, and Hosobuchi, 1975; Richardson, 1987; Shimoji, Hokari, Kano, et al., 1993). A lack of standardization was found in some of the outcome measures (definitions of the quality of pain relief and VAS reductions) (Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999).
No studies were found that compared the efficacy of SCS to DBS for the treatment of pain.
The five studies on SCS were all case series and involved a total of 84 patients. No controlled trials were found. The sample sizes of the studies varied from 12 cases to 23 patients with TSCI. Two studies were prospective (Richardson, 1987; Shimoji, Hokari, Kano, et al., 1993) and three were retrospective (Broggi, Servello, Dones, et al., 1994; Meglio, Cioni, Prezioso, et al., 1989; Nielson, Adams, and Hosobuchi, 1975). The general inclusion criteria for these studies included chronic pain unresponsive to medical therapy, acupuncture, TENS, and physiotherapy. Only one study mentioned any exclusion criteria (major psychiatric illness) (Meglio, Cioni, Prezioso, et al., 1989).
One study permitted extraction of data on the number of patients with SCI at various levels of the spine. The number of patients in this study was 16 (Meglio, Cioni, Prezioso, et al., 1989). Eleven of them had thoracic, four had lumbar, and one patient had a cervical level SCI. Completeness of the injury was not reported or not extractable in any of the five studies.
Descriptive information regarding the gender and age of patients was found in two studies (Meglio, Cioni, Prezioso, et al., 1989; Shimoji, Hokari, Kano, et al., 1993) representing 28 patients. Seventy-four percent were male. Mean ages of this sample were 51 and 54.9 years. No studies provided a definition of neuropathic pain.
Outcome measures included categorical measures of pain relief (Broggi, Servello, Dones, et al., 1994; Nielson, Adams, and Hosobuchi, 1975); percent pain relief (Meglio, Cioni, Prezioso, et al., 1989; Richardson, 1987); binary pain relief: > 50% pain relief (Meglio, Cioni, Prezioso, et al., 1989; Shimoji, Hokari, Kano, et al., 1993), and satisfactory/unsatisfactory (Nielson, Adams, and Hosobuchi, 1975). Reductions in medication usage were used as an outcome in two studies (Meglio, Cioni, Prezioso, et al., 1989; Shimoji, Hokari, Kano, et al., 1993) but were reported only in one of the studies (Shimoji, Hokari, Kano, et al., 1993).
SCS can be achieved through percutaneous insertion of electrodes into the epidural space (PSCS) or via surgical placement of electrodes on the dura above or below the lesion (Buchhaas, Koulousakis, and Nittner, 1989) for dorsal column stimulation (DCS). One study reported on the relatively long-term use of percutaneously-placed electrodes (Shimoji, Hokari, Kano, et al., 1993). The remaining four studies involved patients whose potential responsiveness to SCS was assessed using PSCS prior to surgical implantation or patients receiving DCS (Broggi, Servello, Dones, et al., 1994; Meglio, Cioni, Prezioso, et al., 1989; Nielson, Adams, and Hosobuchi, 1975; Richardson, 1987). One study examined the use of continuous low frequency percutaneous epidural SCS (Shimoji, Hokari, Kano, et al., 1993).
One study reported statistical analysis of the results (Shimoji, Hokari, Kano, et al., 1993). The five studies involving SCS used similar outcome measures. In general, the studies showed initial improvement with stimulation in 50 to 70 percent of selected patients but a decline in effectiveness of stimulation on followup to 19 to 41 percent (Meglio, Cioni, Prezioso, et al., 1989; Richardson, 1987).
Adverse effects of SCS were reported for entire study samples. Information on adverse effects in the TSCI population was not usually extractable. It included lead displacement or breakage and failure of pain relief, cerebrospinal fluid leak, infection of the subcutaneous pocket or infection of the surgical wound, muscle spasms, and radicular muscle twitches. No major complications were noted in any of the studies.
The 12 case reports or small case series showed similar results, with at least one or two patients receiving partial pain relief and some patients having complications resulting from the procedure.
Cumulatively, the six studies examined 21 patients after TSCI (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999; Young, Tronnier, and Rinaldi, 1992). Twenty out of the 21 patients studied had incomplete SCI.
Descriptive information regarding the age and gender of patients with CNP after TSCI was variably reported. Most of the studies included the data in summary form and included patients with chronic pain of differing etiologies, making this information unextractable for patients with TSCI (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Lefaucheur, Decq, et al., 1999). Ages of patients with TSCI were reported for 14 patients in three studies, with a mean age of 47 years (Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997; Young, Tronnier, and Rinaldi, 1992). Gender was reported for three patients in two studies (all were male) (Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997). Age and gender-related data were completely absent in one article (Hosobuchi, Adams, and Rutkin, 1975).
An explicit description of the pain reported by patients was given in three of the eight studies (Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999) and consisted of severe (2/3 studies), constant (2/3 studies), superficial (1/3 studies) or deep (2/3 studies), diffuse (2/3 studies), burning (3/3 studies), or aching (2/3 studies) pain.
The time of pain onset after injury was reported in one report (Levy, Lamb, and Adams, 1987). The length of time that CNP was experienced by patients with TSCI prior to DBS was reported in two studies (Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997) and was 14 years. The mean followup after DBS was reported nonspecifically for the study samples and varied from 27 to 80 months.
Patients with deafferentation pain following TSCI were treated by stimulation of the sensory thalamus (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987), the internal capsule (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987), or the motor cortex (Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999). Exact electrode implantation sites were not extractable in one study (Kumar, Toth, and Nath, 1997). The stimulation parameters, when reported, were noted in general terms for the entire study sample, and the data for patients with TSCI were not extractable (Kumar, Toth, and Nath, 1997; Nguyen, Lefaucheur, Decq, et al., 1999).
The studies cited examined several types of outcome measures. These included binary outcomes for short-term pain relief (Hosobuchi, Adams, and Rutkin, 1975; Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997); long-term pain relief (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987; Nguyen, Keravel, Feve, et al., 1997); percentage reduction in pain (Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999); a modified MPQ (Kumar, Toth, and Nath, 1997), VAS measures of pain pre- and postoperatively (Kumar, Toth, and Nath, 1997; Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999); ordinal measures of improvement in the activities of daily living (Kumar, Toth, and Nath, 1997; Nguyen, Lefaucheur, Decq, et al., 1999); and reduction in medication usage (Levy, Lamb, and Adams, 1987; Nguyen, Lefaucheur, Decq, et al., 1999).
One study reported statistical analysis of the results. The analysis appeared appropriate (Nguyen, Lefaucheur, Decq, et al., 1999). Another reported no statistical difference in outcomes between patients receiving different types of electrodes but did not give further details of the analysis (Kumar, Toth, and Nath, 1997).
The four studies involving DBS used similar outcome measures. One study reported early pain relief in one of three patients with TSCI receiving sensory thalamic or internal capsule stimulation. None of the patients in the study achieved long-term pain relief. The authors believed that failure of the device in the patient with early pain relief was due to the development of tolerance (Kumar, Toth, and Nath, 1997). The followup period was a mean of 78 months for patients in this study.
Stimulation of the periaqueductal gray/periventricular gray matter or the sensory thalamus in patients with SCI led to early pain relief in 4 of 11 patients. None of the patients experienced long-term pain relief (Levy, Lamb, and Adams, 1987). The mean followup for the patients was 3.8 years.
Chronic motor cortex stimulation was found to reduce pain by over 80 percent in one of three patients with paraplegia (Nguyen, Lefaucheur, Decq, et al., 1999). The remaining two patients had 60 percent and 40 percent reductions in pain. Medication use was reduced in the two patients with more optimal responses. In a similar study, chronic stimulation of the motor cortex was reported to produce excellent pain relief (80-100% reduction) in one quadriplegic patient but did not produce pain relief (< 40% reduction) in a paraplegic patient (Nguyen, Keravel, Feve, et al., 1997). Data pertaining to adverse effects of DBS specific to patients with SCI were not extractable. In general, the most common complication of DBS in the studies was headache (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987). Other complications included wound infections (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987); meningitis (Levy, Lamb, and Adams, 1987); fractured electrodes (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987); skin erosions (Levy, Lamb, and Adams, 1987; Nguyen, Lefaucheur, Decq, et al., 1999); blurred vision with stimulation (Kumar, Toth, and Nath, 1997); psychosis (Levy, Lamb, and Adams, 1987); hardware malfunction or migration (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987); temporary hemi- or monoparesis, confusion, lethargy, dysphagia and local pain with stimulation, foreign body reaction (Levy, Lamb, and Adams, 1987); electrical leaks (Kumar, Toth, and Nath, 1997)' postoperative seizures (Kumar, Toth, and Nath, 1997); intracerebral hematoma (Kumar, Toth, and Nath, 1997; Levy, Lamb, and Adams, 1987); extradural hematoma (Nguyen, Keravel, Feve, et al., 1997; Nguyen, Lefaucheur, Decq, et al., 1999); periventricular hemorrhage (Levy, Lamb, and Adams, 1987); and death (Levy, Lamb, and Adams, 1987).
Spinal stimulation has been used in an attempt to help patients with medically intractable pain find relief. Some patients in the case series identified in this evidence report have experienced at least partial relief of pain. However, it is difficult to place much confidence in these findings since the strength of evidence is poor for studies of SCI and DBS in patients with CNP following TSCI. Existing studies are observational in nature, of small sample size, without control or comparison groups, and have incomplete reports. Detailed descriptions of the patients involved in these studies are lacking. In addition, many of the outcomes specific to patients with CNP were not extractable and were not uniformly defined, such as short-term and long-term success of therapy.
The two studies of TNS report conflicting results. This is not surprising given the differences in rigor of study design. Given the paucity of evidence, it does not appear that TNS is effective in reducing the intensity of CNP, but the experience of "unpleasantness" may be reduced if patients have a positive expectation of treatment effectiveness.
The data on SCS suggest that the technique has a variable rate of early success and a lower rate of long-term efficacy. The DBS data reveal a low rate of early success and an even lower long-term success, coupled with significant adverse risks of the procedure. These findings make it difficult to justify the use of either procedure as a method of treating CNP after TSCI. TNS may reduce the sensation of "pain unpleasantness" if patients have positive expectations of treatment effectiveness.
A discussion of the implications for future research is found in Chapter 8.
DREZ lesions are used to ablate neurons that may demonstrate paroxysmal hyperactivity after deafferentation injury (Sampson, Cashman, Nashold Jr., et al., 1995). In 1942, a Lissauer's tractotomy, the first known operative approach to the human DREZ, was performed for pain relief (Hyndman, 1942). This was followed, in 1966, by the development of an experimental pain relieving procedure -- performed in cats -- called radiofrequency DREZ nucleolysis (Kerr, 1966; Yaksh, 1988). Over the years, the procedure was refined until 1979 when Nashold and Ostahl first performed the DREZ microcoagulation for brachial plexus avulsion pain (Nashold Jr. and Ostdahl, 1979). In 1981, Nashold and Bullitt used it for post-traumatic spinal differentiation pain (Nashold Jr. and Bullitt, 1981).
Rawlings and colleagues suggest that the Rexed's layers I through V are the primary target regions of the DREZ procedure (Rawlings, el-Naggar, and Nashold Jr., 1989). They suggest that these regions are the origin of the spinothalamic tract and, following deafferentiation, there appears to be hyperactive neuronal discharges and changes in the relative concentrations of substance P and beta-endorphins in these layers (Rawlings, el-Naggar, and Nashold Jr., 1989). The putative mode of action of the DREZ procedure remains largely unexplained (Richter and Seitz, 1984). However, pain relief from DREZ lesioning may stem from the following three mechanisms (Richter and Seitz, 1984): interruption of ascending pain pathways within dorsal and dorsal-lateral columns, destruction of pain-generating centers in the spinal cord, or
rebalancing of the inhibitory and excitatory inputs within the damaged sensory network.
DREZ lesions can either be made using radiofrequency or laser (Powers, Barbaro, and Levy, 1988). Computer assisted (CA) guidance has recently been incorporated into the radiofrequency DREZ procedure to improve the accuracy of the lesion placement (Edgar, Best, Quail, et al., 1993). With the advance of technology, laser DREZ procedures have been developed to produce smaller and more discrete lesions (Powers, Barbaro, and Levy, 1988). Specific laser types include carbon dioxide (CO2), argon, and Neodymium: yttrium aluminum garnet (Nd:YAG).
This chapter addresses the question posed by the CSCM: "What is the evidence of effectiveness and safety of DREZ lesioning in treating CNP in patients with TSCI?" We have complemented it with a separate brief description, at the end, of the evidence on other surgical procedures.
The specific search strategies and methods for data extraction are summarized in Chapter 2.
Two of the nine studies made indirect comparisons: one studied radiofrequency DREZ procedures and CO2 laser DREZ (Young, 1990), and the other examined three different laser procedures -- CO2, argon, and Nd:YAG (Powers, Barbaro, and Levy, 1988).
Generally, the study quality was poor. Two studies were prospective (Nashold Jr., Vieira, and el-Naggar, 1990; Powers, Barbaro, and Levy, 1988) and five studies were retrospective (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Rath, Seitz, Soliman, et al., 1997; Sampson, Cashman, Nashold Jr., et al., 1995; Young, 1990). In two studies, the direction could not be determined from the published report (Nashold Jr. and Bullitt, 1981; Wiegand and Winkelmuller, 1985). A high risk of selection bias may be present because eight of nine studies did not explicitly report the inclusion/exclusion criteria for subjects (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Nashold Jr. and Bullitt, 1981; Nashold Jr., Vieira, and el-Naggar, 1990; Powers, Barbaro, and Levy, 1988; Rath, Seitz, Soliman, et al., 1997; Wiegand and Winkelmuller, 1985; Young, 1990). In all but one study (Nashold Jr. and Bullitt, 1981), descriptive information about patients with TSCI was either absent (Edgar, Best, Quail, et al., 1993; Powers, Barbaro, and Levy, 1988; Young, 1990); limited (Rath, Seitz, Soliman, et al., 1997; Wiegand and Winkelmuller, 1985); or inextractable from the complete sample (Friedman and Nashold Jr., 1986; Nashold Jr., Vieira, and el-Naggar, 1990; Sampson, Cashman, Nashold Jr., et al., 1995).
The sample sizes varied from 9 to 54 patients, with a median of 20 patients. Cumulatively, the nine studies examined 215 TSCI patients from a total sample of 459 patients exhibiting a wide range of neurological pathologies (e.g., other conditions included cauda equina injury, nerve root avulsions, tumors). A range was used to describe the number of TSCI patients in one study because six patients with avulsion injuries could not be excluded from the sample (Sampson, Cashman, Nashold Jr., et al., 1995).
Eight out of the nine studies reported 100 percent patient followup (Friedman and Nashold Jr., 1986; Nashold Jr. and Bullitt, 1981; Nashold Jr., Vieira, and el-Naggar, 1990; Powers, Barbaro, and Levy, 1988; Rath, Seitz, Soliman, et al., 1997; Sampson, Cashman, Nashold Jr., et al., 1995; Wiegand and Winkelmuller, 1985; Young, 1990). One study (Edgar, Best, Quail, et al., 1993) reported 91 percent followup of the entire sample (n=112, including non-TSCI patients). Therefore, it was not possible to extract whether TSCI patients contributed to those lost at followup. In all studies, the length of followup ranged from the immediate postoperative period to 5.25 years.
Five studies clearly reported the gender and age of the patients (Friedman and Nashold Jr., 1986; Nashold Jr. and Bullitt, 1981; Nashold Jr., Vieira, and el-Naggar, 1990; Rath, Seitz, Soliman, et al., 1997; Wiegand and Winkelmuller, 1985). The mean ages for four of these studies were 47.4, 39.8, 47, and 46 years, respectively. While Freidman reported a range of 27 to 72 years. In addition, in these five studies, male subjects represented 85 percent, 78 percent, 56 percent, 83 percent, and 88 percent of the TSCI sample. The level of injury was not reported in five studies (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Nashold Jr., Vieira, and el-Naggar, 1990; Wiegand and Winkelmuller, 1985; Young, 1990). Of the remaining studies, one examined patients with conus medullaris lesions (Sampson, Cashman, Nashold Jr., et al., 1995). In the other studies, the thoracic region was the most common level of injury (Nashold Jr. and Bullitt, 1981; Powers, Barbaro, and Levy, 1988; Rath, Seitz, Soliman, et al., 1997).
Six studies did not report the completeness of the lesion (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Nashold Jr., Vieira, and el-Naggar, 1990; Powers, Barbaro, and Levy, 1988; Wiegand and Winkelmuller, 1985; Young, 1990). Two studies categorized patients' injuries as complete/incomplete but failed to report the percentages (Rath, Seitz, Soliman, et al., 1997; Sampson, Cashman, Nashold Jr., et al., 1995). The remaining study reported 78 percent of the patients had complete injuries and 22 percent had incomplete motor injuries (Nashold Jr. and Bullitt, 1981). Although all studies failed to clearly report or present extractable data on duration of pain, onset of pain, and time since injury, they did provide descriptors of neuropathic pain. The more common descriptors were "burning" (8/9 studies), "electric," "sharp," and "deafferentation pain" (3/9 studies). Other descriptors are reported in the evidence tables.
Six studies used radiofrequency DREZ (Friedman and Nashold Jr., 1986; Nashold Jr. and Bullitt, 1981; Nashold Jr., Vieira, and el-Naggar, 1990; Rath, Seitz, Soliman, et al., 1997; Sampson, Cashman, Nashold Jr., et al., 1995; Wiegand and Winkelmuller, 1985). One study used two variations of radiofrequency DREZ (temperature controlled and frequency controlled) and CO2 laser DREZ (Young, 1990). Another study used CA radiofrequency DREZ (CA DREZ) (Rath, Seitz, Soliman, et al., 1997), and the final study used three types of laser DREZ (CO2, argon, Nd:YAG) (Powers, Barbaro, and Levy, 1988).
Although all studies used some type of categorical or binary outcome measure, there was variability across studies regarding the names and definitions of outcomes. Three studies used a binary outcome (greater or less than 50 percent pain relief) to report results (Edgar, Best, Quail, et al., 1993) (Powers, Barbaro, and Levy, 1988; Young, 1990), and another study used a binary outcome of success/failure in achieving 100 percent pain relief (Wiegand and Winkelmuller, 1985).
Five studies used a categorical outcome measure (Friedman and Nashold Jr., 1986; Nashold Jr. and Bullitt, 1981; Nashold Jr., Vieira, and el-Naggar, 1990; Rath, Seitz, Soliman, et al., 1997; Sampson, Cashman, Nashold Jr., et al., 1995). Categorical pain outcomes were determined by patients' ratings of pain relief, decreased pain medication usage, and interference with daily activities. None of the studies reported the validity, reliability, and responsiveness of the outcome measures. Furthermore, no study reported blinding of the outcome assessors.
There was no report of statistical analysis performed in any of the studies.
In three radiofrequency DREZ studies (Friedman and Nashold Jr., 1986; Nashold Jr., Vieira, and el-Naggar, 1990; Sampson, Cashman, Nashold Jr., et al., 1995), the authors reported 26-28 of 54 (48-52%), 12 of 16 (75%), and 12-15 of 29 (41-52%) TSCI patients experienced good pain relief 1 year postsurgery, immediately postsurgery, and at a mean followup of 3 years, respectively. Good pain relief was defined as no analgesic use and no limitation of activity by pain. In these same studies, the authors also reported that three to five of 54 (6-9%), four of 16 (25%) and seven of 29 (24%) TSCI patients experienced fair pain relief (nonnarcotic use and no limitation of activity by pain) at the same followup. A range was used to describe the number of TSCI patients in the first study because two patients with spinal tumors could not be excluded from the sample (Friedman and Nashold Jr., 1986).
Another radiofrequency DREZ study reported that eight of 15 (53 percent) patients experienced 100 percent pain relief at an unreported followup (Wiegand and Winkelmuller, 1985). In a similar study, the authors reported that 11 of 23 (47%) TSCI patients had greater than 75 percent pain relief at a mean followup of 51 months after undergoing radiofrequency DREZ (Rath, Seitz, Soliman, et al., 1997). The remaining radiofrequency DREZ study reported greater than 50 percent pain relief in seven of nine (78%) TSCI patients at followups ranging from 5-38 months (Nashold Jr. and Bullitt, 1981).
In the only study using CA DREZ (Edgar, Best, Quail, et al., 1993), the authors reported that 84 percent of TSCI patients experienced 100 percent pain relief, and 92 percent experienced 50-100 percent pain relief at a mean followup of 44 months.
In the only study reporting the effectiveness of radiofrequency DREZ and CO2 laser DREZ, satisfactory pain relief (defined as pain reduction of at least 50%, cessation of narcotic use, and improvement in functional capacity) was found in eight of 15 (53 percent) TSCI patients undergoing radiofrequency DREZ and three of six (50%) undergoing laser DREZ (Young, 1990). The followup range of the entire sample (n=78) was 3.0-5.1 years.
One study examined the use of three different types of laser DREZ (CO2, argon, Nd:YAG) (Powers, Barbaro, and Levy, 1988). The results for TSCI patients were reported in aggregate across the three laser types. The authors reported that five of nine TSCI patients experienced greater than 50 percent pain relief with no narcotic use at a mean followup of 24 months.
Two studies, using radiofrequency DREZ, did not report the assessment of adverse effects (Nashold Jr., Vieira, and el-Naggar, 1990; Wiegand and Winkelmuller, 1985). Four studies reported adverse effects for TSCI subjects only (Nashold Jr. and Bullitt, 1981; Powers, Barbaro, and Levy, 1988; Sampson, Cashman, Nashold Jr., et al., 1995; Young, 1990). An adverse effect common in all four studies was motor weakness. This was observed in 14 of 61-67 patients. Cerebral spinal fluid leak and wound infections were reported in two of these studies (Powers, Barbaro, and Levy, 1988; Sampson, Cashman, Nashold Jr., et al., 1995). Each of these adverse effects was observed in 3 of 32-38 patients, and the severity was not reported. As indicated previously, a range was used to describe the number of TSCI patients in one study because six patients with avulsion injuries could not be excluded from the sample (Sampson, Cashman, Nashold Jr., et al., 1995). Three studies reported adverse effects for the entire sample (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Rath, Seitz, Soliman, et al., 1997). Although there was greater variability in the adverse effects reported by these studies, motor and sensory deficits remained the most commonly reported adverse effects (Edgar, Best, Quail, et al., 1993; Friedman and Nashold Jr., 1986; Rath, Seitz, Soliman, et al., 1997).
Cumulatively, the six case series described 157 subjects. The types of surgical interventions included the following: anterior depression (1 study); spinal cord untethering (1); cordectomy (1); and a variety of interventions such as cordotomy, myelotomy, and arachnoid grafting (3). Across the six case series, improvement in pain varied from 20 percent to 85 percent in selected patients. Specific data related to duration of pain and onset of pain were not reported in five of six studies. Adverse effects included serious complications such as death due to pulmonary embolism (one patient), kyphosis, cerebrospinal fluid leak, meningitis, and refracture of the spine at another level.
The nine studies reviewed indicate that DREZ lesioning in the treatment of CNP may offer promising results. Regardless of type of DREZ procedure, this intervention seems to have favorable pain relief results for more than half of the patients. As DREZ lesioning has been successfully used to treat neuropathic pain resulting from avulsion injuries, these results for CNP must be interpreted carefully. Given the traumatic nature of the injuries experienced by patients included in this review, it must be a difficult task for investigators to effectively rule out the presence of peripheral pathology in their patients. Therefore, it is possible that some of the patients described in the studies as having central pathology may as well have had concurrent peripheral pathology. Therefore, even as this review attempted to exclude patients with peripheral pathologies such as avulsion injuries, the positive results of DREZ lesioning may reflect the successful reduction of neuropathic pain due to peripheral pathology.
More importantly, the strength of evidence provided by the nine DREZ lesioning studies and the four studies of other surgical procedures reviewed was poor. The studies were observational in nature and did not use a control group (through either random allocation or matching on relevant variables) or blinded assessments. These shortcomings lead to the potential for confounded or invalid results. For this reason, it is uncertain whether improvement in patients' CNP should be attributed to chance, natural progression of the pain, or to the effectiveness of the DREZ lesioning or the other spinal surgical procedures.
As well, all of the studies on DREZ lesioning and other spinal surgeries had poorly defined, or lacked, inclusion and exclusion criteria for their samples. Moreover, there were insufficient descriptions of patient characteristics. Therefore, it is not easy to generalize the results of these studies to other groups of TSCI patients with CNP.
Even recognizing the problems regarding the validity and generalizability of the studies, some may look to DREZ lesioning or other spinal surgeries as a last resort when other palliative efforts have failed. Given that the studies did not adequately report the severity of the adverse effects experienced by patients, it is unknown whether DREZ lesioning and other spinal surgeries pose unwarranted risks to patients. For these reasons, even in situations where previous interventions have failed, the evidence is weak for the support of DREZ lesioning or other spinal surgeries to relieve CNP in TSCI patients.
A discussion of the implications for future research is found in Chapter 8.
Central neuropathic pain is a significant problem experienced by many people after a TSCI. Most estimates of the prevalence of chronic pain after TSCI range from 40 percent to 75 percent of patients. Pain is moderate to severe in 25 percent to 60 percent of these individuals, is often associated with psychological and psychiatric conditions, and is severe enough to impair or prevent optimal physical function and daily living. Given the extent and burden of the problem, it was disappointing to find relatively little research on important areas such as the assessment or treatment of CNP after TSCI. Perhaps the state of research is understandable given the complexity of the condition; the variety of plausible but poorly understood underlying mechanisms of CNP; and major difficulties (ethical, logistical, and methodological) in conducting research in this area resulting in tentative and incomplete data.
This Evidence Report has summarized the available research studies, thereby setting the stage for the development of a research agenda. This agenda needs to include methods to improve the quality of research designs; since nearly all of the research published to date is of poor methodological quality, with most intervention studies lacking a control group, blind assessment of the primary outcomes, and adequate followup of subjects. There are also problems related to generalizability that may reflect poor reporting or, more likely, indicate weak study execution. These problems include lack of explicit inclusion criteria; missing or incomplete demographic descriptions of the study sample; and missing or incomplete information related to the severity, location, quality, and nature of the pain.
Given the state of the body of existing evidence and the mandate of AHRQ, this final chapter focuses on the implications for further research to assist in the development of a research agenda rather than on general implications for clinical practice. Specific implications for clinical practice are found in each chapter. It is imperative to develop effective strategies to improve the number, validity, precision, and relevance of future studies.
Future research efforts should consider:
Multicenter collaboration to set a research agenda. The CSCM may be well positioned to facilitate this level of collaboration, or alternative strategies may be needed to foster pragmatic working relationships, even among groups that do not have a tradition of cooperation. Such collaborative groups could study the research problems and provide training in clinical research to young investigators.
Developing of standardized definitions of CNP and a core set of outcome measures. Outcomes should be important to patients, clinicians, and purchasers (e.g., proportion of patients achieving adequate analgesia, impact of treatment on quality of life, unacceptable adverse effects, resource utilization).
Larger studies involving multiple centers (including international collaboration) with more rigorous design, more comprehensive reports, and longer term followup are needed to establish the effectiveness and adverse effects of most of the interventions available. Special emphasis should be placed on gathering evidence on the effects of different interventions in women and adolescents.
Focus on studies of people with TSCI and CNP rather than including individuals poststroke or with peripheral nerve injuries.
Priority should be given to interventions with established roles for the management of other types of neuropathic pain, such as tricyclic antidepressants, anticonvulsants, local anesthetics, and opioids. Studies designed to judge the added value of these interventions given in combination (including treatment algorithms), through invasive routes (e.g., epidural and intrathecal infusions of opioids and local anesthetics), or using different formulations (e.g., sustained release preparations) should also be a priority.
Since CNP is associated with psychosocial difficulties, other noninvasive approaches such as multidisciplinary or self-management approaches should be developed and evaluated for those with TSCI.
Based on the evidence available, SCS and DREZ lesions may be useful in selected individuals. These interventions, however, are also invasive and potentially harmful. The studies that are needed will require complex, controlled designs with close attention to safety issues, substantial amounts of resources, and efficient collaboration among research groups.
Studies are also needed to determine whether the response to treatment is influenced by the level and cause of the SCI as well as by the duration, distribution, and characteristics of the pain, and by comorbid factors (e.g., anxiety and depressive disorders).
There is a great opportunity for consumer groups to call for and support more research activities, given the number of important questions that remain unanswered.
Funding and conducting the research that is required will not be easy, given the complexity of the disorder, the frequent presence of comorbidity, and the variety of interventions and outcomes available.
Most of the current problems we encountered could be easily corrected if journal editors adopted evidence-based reporting recommendations such as the Consolidation of the Standards of Reporting Trials (CONSORT) statement (Begg, Cho, Eastwood et al., 1996) and kept track of new methodological developments that could increase the validity and applicability of research. The CONSORT statement was produced and published by an international group of clinical epidemiologists, biostatisticians, and journal editors in 1996 and is now available on the Internet (www.icmje.org). The aim of this statement is to improve the standards of written reports of RCTs and to ensure that readers find all the information they require in the reports to interpret the trial results with confidence. This statement includes a checklist of 21 items and a flow diagram that authors can use to provide necessary information on the progress of patients through a study. The statement has already been adopted by over 70 major biomedical journals (Jadad and Rennie, 1998). Similar efforts are evolving in relation to observational studies.
The findings and conclusions of this Task Order are based on the information that was available in the published reports of the studies included. Additional information obtained directly from the authors could have overcome many of the reporting limitations described above. Contact with authors could have also led to reduction in the likelihood of publication bias through the identification of unpublished studies. The budget and timelines available, however, were insufficient to allow this.
The interpretability of the data included in most of the tables of this evidence report is limited because so many different outcome instruments were used, often with limited descriptions.
In summary, this report includes the first set of systematic reviews on the management of CNP following TSCI. They incorporate state-of-the-art methodology and are ready for incorporation into evidence-based clinical practice guidelines or performance measures. The report also provides a detailed description of the many limitations of the evidence available and provides recommendations to fill existing knowledge gaps through rigorous research. Filling such gaps will not be easy and will require highly innovative efforts and collaboration among different groups of decisionmakers. If this field continues to produce few, small, incompletely reported studies with heterogeneous designs instead of the rigorous and ambitious high-quality collaborative efforts required, research in this area will continue to be of little value to guide important clinical and policy decisions.
Citations of Articles Not Located
Akman MN, Ersoy Y. Chronic pain and its management in spinal cord injured patients. Fizik Tedavi Rehabilitasyon Dergisi 1995;3.
Ammer K, Rathkolb O. Biofeedback treatment in neurological and head diseases -- Literature survey. Rehabilitacia. Supplementum 1997;30(4):237-42.
Kowaluk EA, Arneric SP, Williams M. Opportunities in pain therapy: Beyond the opioids and NSAIDs. Emerg Drugs 1998;3:1-37.
Ozyurt M, Hizmetli S, Nacitarhan V, Elden H, Kant B, Goker I. The relationship between different variables and prognosis in paraplegic patients. Fizik Tedavi Rehabilitasyon Dergisi 1997;21(1):16-9.
Studies of Pharmacological Interventions
The Consortium of Spinal Cord Medicine (CSCM), a group of 18 health professional organizations representing physicians, therapists, nurses, psychologists, and social workers, organized and supported by PVA, develops and disseminates clinical practice guidelines to professionals and consumers.
American Academy of Orthopedic Surgeons
American Academy of Physical Medicine and Rehabilitation
American Association of Neurological Surgeons
American Association of Spinal Cord Injury Nurses
American Association of Spinal Cord Injury Psychologists and Social Workers
American Congress of Rehabilitation Medicine
American Occupational Therapy Association
American Paraplegia Society
American Physical Therapy Association
American Psychological Association
American Spinal Injury Association
Association of Academic Physiatrists
Association of Rehabilitation Nurses
Congress of Neurological Surgeons
Eastern Paralyzed Veterans Association
Insurance Rehabilitation Study Group
Paralyzed Veterans of America
U.S. Department of Veterans Affairs
Alejandro (Alex) Jadad, MD, DPhil (Task Order Leader): Dr. Jadad is a Professor in the Department of Clinical Epidemiology and Biostatistics, Director of the McMaster University Evidence-based Practice Centre, Chief of the Health Information Research Unit, Co-Director of the Canadian Cochrane Centre at McMaster University in Hamilton, Ontario, Canada. His clinical training is in anesthesiology, pain relief, and palliative care. Dr. Jadad has a Doctor of Philosophy degree in systematic reviews and meta-analyses in pain relief from the University of Oxford (Balliol College).
Dr. Jadad has extensive experience in the study and management of neuropathic pain. He has participated in RCTs on the role of opioids, NMDA receptor antagonists, and alpha-2-agonists during the management of neuropathic pain. He has also conducted systematic reviews on the effect of opioids, antidepressants, sympathetic blockades and anticonvulsants for the relief of neuropathic pain and reviewed the evidence on the prevalence and management of pain in long-term care facilities. In 1993, he was invited to present the best available knowledge on the management of neuropathic pain at the 7th World Congress on Pain in Paris. He discussed the role of systematic reviews in pain relief and presented the database on the measurement of pain at the 8th World Congress on Pain in Vancouver in 1996. He will lead the first refresher course on evidence-based decisionmaking at a World Congress on Pain in Vienna this year.
Dr. Jadad has also developed innovative methods to include consumers as members of the research team. His methods have enabled the development of the first validated tool to assess the quality of RCTs, the largest databases of RCTs and measurement tools in pain relief, the guide "A Team Approach to Pain Relief," tools for the development of consumer versions of clinical practice guidelines, questionnaires for the Internet surveys, and projects to improve the treatment of pain in nursing homes and the measurement of pain in general.
Mary Ann O'Brien, MSc (Task Order Coordinator): Ms. O'Brien is an Assistant Clinical Professor in the School of Rehabilitation Sciences at McMaster University. She has graduate training in the science of conducting systematic reviews and is a licensed physical therapist with extensive clinical experience in rehabilitation. She is a member of the Board of Examiners for the Canadian Physical Therapy Examination. She is also a member of the Cochrane Collaboration Effective Professional Practice and Organization of Care Review Group. Ms. O'Brien has extensive experience in coordinating the production of systematic reviews. Recently, she coordinated the production of 13 systematic reviews of public health interventions for the Ontario government. In addition, she is the author or coauthor of 10 systematic reviews in the areas of health professional behavior change, public health, and temporomandibular joint disorders.
Charles Goldsmith, PhD (Biostatistics): Dr. Goldsmith is a Professor and statistician in the Department of Clinical Epidemiology and Biostatistics and Associate Member of the Department of Mathematics and Statistics, McMaster University; and Head of Biostatistics, Father Sean O'Sullivan Research Centre, St. Joseph's Health Care System Research Network. He will lead and supervise the statistical aspects of the project. He is interested in the measurement of the attitudes of the public and in the study of the validity of diagnostic tests and has provided statistical input to the validation of measurement tools to assess the methodological quality of review articles and the functional status of patients with juvenile arthritis. He received the 1995 Founders Award of the American Statistical Association. In 1992 and 1998, he received the Silver Quill Award for Basic and Applied Research for his work on observer variation in an audit of charts of patients with rheumatoid arthritis and on the measurement of lumbar flexion, respectively.
Ann McKibbon, MLS (Library Sciences): Mrs. McKibbon is an Associate Professor in the Department of Clinical Epidemiology and Biostatistics. She is well known internationally for her expertise in searching and reviewing the medical literature and supervises the production of several evidence-based journals produced at McMaster, including Evidence-based Medicine, Evidence-based Mental Health, Evidence-based Nursing, Best Evidence, and ACP Journal Club. She implemented the first Canadian demonstration site of an online searching system for the Canadian Institute of Scientific and Technical Institute in 1973. Since then, she has collaborated on and led many research projects related to information retrieval. She has taught more than 2,000 physicians MEDLINE-searching skills and more than 1,000 health science librarians the basic principles of clinical epidemiology and how they apply to information retrieval. Her research on clinical informatics is published widely and used extensively by clinicians, teachers, researchers, and systematic reviewers. She is the author of the book, "PDQ Evidence-based Principles and Practice," published in 1999.
Gordon Guyatt, MD, FRCPC, MSc (Measurement Methodology): Dr. Guyatt is a Professor in the Departments of Medicine, and Clinical Epidemiology and Biostatistics, McMaster University. He is the person who coined the term "Evidence-based Medicine." He has written many systematic reviews and meta-analyses of different clinical topics in medicine and nursing on diagnosis, measurement, prognosis, and treatment. He has authored several landmark methods papers addressing key issues in measurement, RCTs, systematic reviews, and levels of evidence. He is Chair of the Evidence-based Medicine Working Group. From this role, he has written and/or edited the series of articles in JAMA called the Users' Guides to the Medical Literature, translated now into 11 languages.
Bernie O'Brien, PhD (Health Economics): Dr. O'Brien is an Associate Professor in the Department of Clinical Epidemiology and Biostatistics, Associate member of the Centre for Health Economics and Policy Analysis at McMaster University, and Associate Director of the Centre for Evaluation of Medicines at St. Joseph's Hospital. His work focuses on methods of economic evaluation as applied to new therapeutic interventions, particularly in the context of clinical trials. Other interests include decision analysis, measurement of health-related quality of life, pharmaceutical utilization, and risk assessment. He is a recipient of a PMAC Health Research Foundation -- Medical Research Council of Canada Career Award in Health Sciences (1995-2000).
Dean Wingerchuk, MD: Dr. Wingerchuk received his MD from the University of Saskatchewan. He performed his internship, neurology residency, and clinical neuroimmunology fellowship at the Mayo Clinic in Rochester, MN. Subsequently he has completed a multiple sclerosis fellowship at the University of Western Ontario and the MSc program in Health Research Methodology at McMaster University. As of July 2000, Dr. Wingerchuk will be the senior associate consultant (Mayo Clinic, Scottsdale, AZ) and Assistant Professor of Neurology (Mayo Medical School, Rochester, MN). His clinical and research interests are multiple sclerosis (MS), natural history of MS, and the use of databases for observational and experimental studies in MS.
Pamela Angle, MD, DABA, FRCP (C): Dr. Angle is an Assistant Professor at the Department of Anesthesia at Sunnybrook and Women's College Health Sciences Centre in Toronto. She is completing her Masters in Health Research Methodology at McMaster Univesity. Her special interests include obstetrical anesthesia.
Julian Mulcaster, MD, FRCP (C): Dr. Mulcaster is a Clinical Scholar, Pain Management, Department of Anaesthesia, McMaster University and Hamilton Health Sciences Corporation.
Carmen Tamayo, MD: Dr. Tamayo has an MD Degree from the Central University of Venezuela and a degree in Public Health from the University of Toronto. She has acted as a research advisor for the University of Texas Center for Alternative Medicine Research in Cancer (UT-CAM), a National Institute of Health (NIH)-funded exploratory research center associated with the National Cancer Institute and the National Center for Complementary and Alternative Medicine in the United States. She is currently director of the Complementary and Alternative Medicine Division of Foresight Links Corporation, a private consultanting firm engaged in data analysis and evidence-based medicine. She is involved in research development, clinical protocol design, and review of CAM proposals and related literature, developing and coordinating scientific protocols to assess the effectiveness of CAM therapies in clinical trials. She is presently involved with the Health Information Research Unit and the Canadian Center for Evidence-based Medicine at McMaster University in Hamilton, Ontario. Dr. Tamayo is a member of the College of Physicians in Venezuela, the Drug Information Association, the National Breast Cancer Coalition, and the Canadian Cochrane Collaboration.
Matthew Denkers: Mr. Denkers graduated in 1998 from McMaster University with a Bachelor's degree in Kinesiology with a Minor in Biology. He graduated from McMaster in September 2000 with a Bachelor of Health Sciences Degree (Physiotherapy). He has entered Medical School at McMaster.
Heather Biagi: Ms. Biagi completed a Bachelor of Physical Education (BPE) degree at the University of British Columbia and a Masters of Science (MSc) degree at the University of British Columbia where her area of research was on the effectiveness of bracing on knee osteoarthritis. She graduated in September 2000 as a Physiotherapist BHSc (PT). Her area of interest is Orthopedics/Sports Medicine.
Mary Gauld: Ms. Gauld is a Research Coordinator in the Department of Clinical Epidemiology and Biostatistics, McMaster University. She has participated in numerous systematic reviews, including the AHRQ Task Order on the Treatment of Attention-Deficit/Hyperactivity Disorder.
Alexia Antzack-Bouckoms, DMD ScD MPH: Dr. Antzack-Bouckoms is the Co-Director of the New England Cochrane Centre and one of the Editors of the Cochrane Oral Health Group. She is a dentist with vast experience in systematic review methodology. She suffered spinal cord injury, has neuropathic pain, and has become a very active consumer representative, raising awareness about the importance of rigorous research on spinal cord injury.
Gary Bennett, PhD: Dr. Bennett is Professor, Department of Neurology, Director of Pain Research, and Director of the Quantitative Sensory Testing and Autonomic Nervous System Testing Laboratories, MCP Hahnemann University, Philadelphia, PA. He acted as Head, Neuropathic Pain and Pain Assessment Group, Neurobiology and Anesthesiology Branch, National Institute of Dental Research, NIH, Bethesda, MD, from 1990 to 1991, and as Chief, Neuropathic Pain and Pain Measurement Section, Neurobiology and Anesthesiology Branch, National Institute of Dental Research, NIH, Bethesda, MD, from 1991 to 1996. Dr. Bennett is one of the most productive basic researchers in the field of neuropathic pain in the world.
Daniel Carr, MD: Dr. Carr is Saltonstall Professor of Pain Research; Vice Chair for Research, Department of Anesthesiology, Tufts-New England Medical Center, Boston, MA. He has background and expertise in internal medicine, anesthesiology, pain relief, and basic research. He has an extensive research record in pain relief. From 1990 to 1992, Dr. Carr acted as Co-Chair of the USDHHS/AHCPR, Acute and Cancer Pain Management Guideline Panels. He is currently the lead editor within the Cochrane Pain, Palliative and Supportive Care (PaPas) Collaborative Review Group.
John Farrar, MD: Dr. Farrar is Adjunct Assistant Professor of Epidemiology and Anesthesia; and Senior Scholar, Center for Clinical Epidemiology & Biostatistics, University of Pennsylvania, Philadelphia, PA. He is a neurologist with strong methodological expertise, specifically on clinical and methodological issues related to the definition, diagnosis, and management of neuropathic pain.
Angela Mailis, MD, MSc, FRCPC: Dr. Mailis is Director, Comprehensive Pain Program and Pain Investigation Unit, The Toronto Hospital; Associate Professor, Department of Medicine, and Associate member of the Playfair Neuroscience Unit, University of Toronto, Canada. She is also Co-chair of the College of Physicians and Surgeons of Ontario Chronic Pain Task Force. Dr. Mailis is a physiatrist with a Master of Science degree in human spinal plasticity. She has spent most of her research career studying human neuropathic pain, human pain syndromes, the effect of demographics on pain, and adverse effects of invasive procedures for pain management. She is currently working on several systematic reviews as a member of the Cochrane PaPas Group.
Dwight Moulin, MD: Dr. Moulin is an Associate Professor, Clinical Neurological Sciences and Oncology, at the University of Western Ontario, London, Ontario. He is a member of the College of Physicians and Surgeons of Ontario Task Force on Management of Chronic Nonmalignant Pain. His main clinical and research interests are in pharmacological and nonpharmacological treatment of neuropathic pain.
Kenneth Parsons, MD: Dr. Parsons has served as the chairperson of the Consortium for Spinal Cord Medicine since its inception. He is the Director of the Spinal Cord Injury Program at the Institute for Rehabilitation and Research in Houston, TX, Vice-Chair of the Department of Physical Medicine and Rehabilitation at the University of Texas Medical School, as well as director of the residency program there. He is currently the President of the American Spinal Injury Association. Dr. Parsons' clinical activities frequently include the diagnosis and management of pain states in patients with SCI.
Russell K. Portenoy, MD: Chairman, Department of Pain Medicine and Palliative Care, Beth Israel Medical Center; Professor, Department of Neurology, Albert Einstein College of Medicine. Dr. Portenoy is the current President of the American Pain Society and Editor-in-Chief of the Journal of Pain and Symptom Management. He is a neurologist, has received multiple awards, and is the author of hundreds of articles on pain and its management.
Diana Rintala, PhD: Dr. Rintala is a Research Health Science Specialist at the Center of Excellence on Healthy Aging with a Disability at the Veterans Affairs Medical Center in Houston, TX. She has an appointment as an Associate Professor in the Department of Physical Medicine and Rehabilitation at the Baylor College of Medicine. She has a PhD with a minor in data analytic techniques, with a specific interest in the instruments used to measure pain and the evaluation of the cost implications of different therapies for neuropathic pain.
Christine N. Sang, MD, MPH: Dr. Sang is currently Director, Clinical Trials Program, and Assistant Professor of Anesthesia and Critical Care, Massachusetts General Hospital, Harvard Medical School. Her background is in anesthesiology, epidemiology, and biostatistics. The primary aim of her ongoing research is to study mechanisms of central and peripheral neuropathic pain, particularly after spinal cord injury, and the development and evaluation of dysesthetic pain following spinal cord injury. She currently holds a 5-year grant from the Paralyzed Veterans of America/Spinal Cord Research Foundation for clinical trials evaluating several pharmacological interventions and testing methods for pain after spinal cord injury.
Peter Tugwell, MD: Dr. Tugwell is Physician-in-Chief, Professor and Chairman, Department of Medicine, University of Ottawa; and Chair, Editorial Group of the Cochrane Collaborative Review Group for Musculoskeletal Disease. Dr. Tugwell has received many awards for his work on rheumatology and clinical epidemiology and has hundreds of publications in these areas.
Eldon Tunks, MD FRCP (C): Dr. Tunks is Professor, Department of Psychiatry, McMaster University; Director, Pain Program of the Hamilton Health Sciences Corporation; Co-Chair, College of Physicians and Surgeons of Ontario Task Force on the Management of Chronic Non-malignant Pain. Dr. Tunks was one of the founding members of the International Association for the Study of Pain and is the Associate Editor for Psychiatry for Pain Research and Management, the journal of the Canadian Pain Society. Dr. Tunks is actively involved in the development of evidence-based clinical practice guidelines on the management of chronic nonmalignant pain and has led the critical appraisal of all existing guidelines on this subject produced in North America.
Patricia Huston, MD, MPH. Scientific Communications International, Inc., Ottawa, Ontario
Willem JJ Assendelft MD, PhD. Department of General Practice, Division Public Health and Dutch Cochrane Centre, Academic Medical Centre -- University of Amsterdam, The Netherlands
Michael Cousins, MD. Head, Department of Anaesthesia and Pain Management, University of Sydney, Australia
Fred Cowell. Staff Director, Paralyzed Veterans of America (PVA), Policy Department, Washington, DC
Matthew Elrod, PT, MEd. National Rehabilitation Hospital, American Physical Therapy Association (APTA), VA
Allan Gordon, MD. Associate Professor, University of Toronto, Neurologist Director, Wasser Pain Centre, Mount Sinai Hospital, Toronto, Canada
Neil A. Hagen M.D. Associate Professor, Departments of Oncology, Clinical Neurosciences and Medicine, University of Calgary; and Head, Cancer Pain Clinic, Tom Baker Cancer Centre, Calgary
Claire E. Hulsebosch, Ph.D. Professor, Anatomy and Neurosciences, University of Texas Medical Branch, Galveston, TX
Douglas Justins, M.D. Consultant in Pain Management and Anaesthesia, St. Thomas Hospital, London, United Kingdom
Paul G. Loubser, M.D. Assistant Professor, Department of Anesthesiology, Baylor College of Medicine, Houston, TX
Anthony J. Mariano, Ph.D. Codirector, Pain Clinic, Clinical Psychologist, Spinal Cord Injury Service, Puget Sound Veterans Adminstration Health System, Seattle, WA
Harold Merskey, M.D. Professor Emeritus, Department of Psychiatry, University of Western Ontario, London, Ontario, Canada
Louis Papastrat, MBA. Vice President, Medical Management for American Re-Insurance Company, Princeton, NJ
Daniel Rohe, Ph.D. Associate Professor of Psychology, Mayo Medical School, MN
Audrey J. Schmerzler, RN, MSN, CRRN. Mount Sinai Spinal Cord Injury Model System, Mount Sinai School of Medicine, New York, NY
Paul Shekelle, M.D., Ph.D. RAND Health Program and the Veterans Affairs Health Services Research and Development Service, CA
Richard A. Sherman, PhD. Consultant for Orthopedic Research, Madigan Army Medical Center, Tacoma, WA
Ronald Tasker, MD. Toronto Western Hospital, Toronto, Canada
Frances M. Weaver, PhD. Deputy Director, Midwest Center for Health Services & Policy Research, Hines VA Hospital and Research Associate Professor, Northwestern University, Hines, IL
Dawn M. Sexton, Project Administrator for CPGs, Consortium for Spinal Cord Medicine, Paralyzed Veterans of America, Washington, DC
J. Paul Thomas, Consortium Coordinator, Consortium for Spinal Cord Medicine, Paralyzed Veterans of America, Washington, DC
| Database: Medline <1966 - May 2000> Set Search | Results | |
| 001 | exp spinal cord injuries/ | 16264 |
| 002 | spinal cord injur:.tw. | 7297 |
| 003 | spinal cord trauma:.tw. | 442 |
| 004 | spinal cord compression.tw. | 1674 |
| 005 | parapleg:.tw. | 5819 |
| 006 | quadripleg:.tw. | 1867 |
| 007 | tetrapleg:.tw. | 1392 |
| 008 | exp paraplegia/ | 6776 |
| 009 | quadriplegia/ | 4159 |
| 010 | neurologic: deterioration.tw. | 1207 |
| 011 | exp spinal cord/ | 46431 |
| 012 | exp wounds/ | 331681 |
| 013 | 11 and 12 | 3004 |
| 014 | 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 or 13 | 28974 |
| 015 | nociception.tw. | 1891 |
| 016 | (anaesthesia dolorosa or anesthesia dolorosa).tw. | 60 |
| 017 | hyperalgesia.tw. | 1854 |
| 018 | causalgia.tw. | 262 |
| 019 | exp hyperesthesia/ | 1399 |
| 020 | (hyperaesthesia or hyperesthesia).tw. | 274 |
| 021 | hypersensitivit:.tw. | 23726 |
| 022 | allodynia.tw. | 626 |
| 023 | causalgia/ | 318 |
| 024 | (post traumatic pain or posttraumatic pain).tw. | 78 |
| 025 | (dysesthe: pain or dysaesthe: pain).tw. | 41 |
| 026 | neuropath: pain.tw. | 903 |
| 027 | neurogen: pain.tw. | 114 |
| 028 | (phantom adj4 pain).tw. | 390 |
| 029 | paraplegic pain.tw. | 8 |
| 030 | central pain.tw. | 264 |
| 031 | chronic pain.tw. | 4576 |
| 032 | sci pain.tw. | 10 |
| 033 | deafferent:.tw. | 3069 |
| 034 | (central dysesthe: or central dysaesthe:).tw. | 5 |
| 035 | (pain adj4 spinal cord).tw. | 555 |
| 036 | (pain adj4 parapleg:).tw. | 114 |
| 037 | (spinal cord adj4 pain).tw. | 555 |
| 038 | or/15-34 | 36947 |
| 039 | 38 and 14 | 337 |
| 040 | 35 or 36 or 37 or 39 | 881 |
| 041 | limit 40 to human | 648 |
| 042 | 41 | 648 |
| Is the purpose of the study to (check all that apply): | ||||
| 1. Assess or diagnose central neuropathic pain? | Yes | No | Not clear | Not reported |
| If yes, go to Section II | ||||
| 2. Validate an outcome tool/measure related to the experience of pain? | Yes | No | Not clear | Not reported |
| If yes, go to Section II | ||||
| 3. Determine the factors that are associated with or could predict the development of chronic central neuropathic pain? | Yes | No | Not clear | Not reported |
| If yes, go to Section III | ||||
| 4. Estimate the prevalence of acute or chronic central neuropathic pain? | Yes | No | Not clear | Not reported |
| If yes, go to Section IV | ||||
| 5. Generate/test a hypothesis related to central neuropathic pain | Yes | No | Not clear | Not reported |
| If yes, stop | ||||
| If no to all questions, describe the purpose of the study: | ________________________________________ | |||
| If yes to question 5 only, stop. | ||||
| If yes to any of questions 1-4, also complete the General Data Extraction Form. | ||||
| Ref ID#_________ | |
| Reviewers Initials_________ |
| Assessment/Outcome Tool/Measure Code___________________________________________________________ | ||||
| 6. Does the study describe the use of the tool in patients with traumatic spinal cord injury (TSCI)? | Yes | No | Not clear | Not reported |
| 7. Does the sample include an appropriate spectrum of individuals with mild and severe pain plus individuals with different but commonly confused disorders, e.g., cauda equina? | Yes | No | Not clear | Not reported |
| 8. Was the setting for this evaluation, as well as the filter through which study patients passed, adequately described? | Yes | No | Not clear | Not reported |
| 9. Was validity of the tool assessed in the study? If yes, how was validity assessed? content construct predictive concurrent other _________________________ Describe the results of the validity assessment: | Yes | No | Not clear | Not reported |
| If no, was validity referenced? | Yes | No | ||
| 10. Was the reproducibility of the test determined in the study? If yes, describe the results. | Yes | No | Not clear | Not reported |
| If no, was reproducibility referenced? | Yes | No | ||
| 11. Was inter-rater variation of the test result determined? If yes, describe the results. | Yes | No | Not clear | Not reported |
| If no, was inter-rater variation referenced? | Yes | No | ||
| 12. Was intra-rater variation of the test result determined (if appropriate)? If yes, describe the results. | Yes | No | Not clear | Not reported |
| If no, was intra-rater variation referenced? | Yes | No | ||
| Ref ID#_________ | |
| Reviewers Initials_________ |
| The next question pertains to studies about the validation of an outcome measure/tool only. | ||||
| 13. Was responsiveness to change measured in the study? If yes, describe the results. | Yes | No | Not clear | Not reported |
| If no, was responsiveness to change reported elsewhere? | Yes | No | ||
| 14. If the test is advocated as part of a cluster or sequence of tests, has its individual contribution to the overall validity of the cluster or sequence been determined? | Yes | No | Not clear | Not reported |
| 15. Have the tactics for carrying out the test been described in sufficient detail to permit their exact replication? | Yes | No | Not clear | Not reported |
| 16. Were any adverse effects reported as a result of using the tool? If yes, describe the adverse effect. | Yes | No | ||
| 17. Briefly, describe any other results of the study | ||||
Complete Questions 18 to 23 for studies where outcomes were measured at more than one point in time.
Complete Questions 21 to 24 if the outcomes were measured at one point in time.
| 18. Was an inception cohort assembled? | Yes | No | Not clear | Not reported |
| 19. Was the referral pattern described? | Yes | No | Not clear | Not reported |
| 20. Was complete follow-up achieved? If no, indicate the follow-up ___________________________ | Yes | No | Not clear | Not reported |
| 21. Were objective outcome criteria developed and used? | Yes | No | Not clear | Not reported |
| 22. Was the outcome assessment blind? | Yes | No | Not clear | Not reported |
| 23. Was adjustment for extraneous prognostic factors carried out? | Yes | No | Not clear | Not reported |
| List the factors that were considered. | ||||
| 24. Briefly describe the results. | ||||
| Ref ID#_________ | |
| Reviewers Initials_________ |
| For studies where the outcomes were measured at one point in time, e.g., a cross sectional survey, also complete questions 25 to 30. | ||||||||||||||||||||||||||||||||||
| 25. How was the sample selected? random consecutive patients other not reported | ||||||||||||||||||||||||||||||||||
| 26. What was the sampling frame? institution community other not reported | ||||||||||||||||||||||||||||||||||
| 27. Was the survey instrument pre-tested? yes no not reported | ||||||||||||||||||||||||||||||||||
28. If the survey instrument was pre-tested, was the instrument:
| ||||||||||||||||||||||||||||||||||
| 29. What was the response rate to the survey? 80-100% 60-79% less than 50% not reported | ||||||||||||||||||||||||||||||||||
| 30. Did the authors address specific hypotheses to be tested? | ||||||||||||||||||||||||||||||||||
| Ref ID#_________ | |
| Reviewers Initials_________ |
| 31. How was the sample selected? random consecutive patients other not reported |
| 32. What was the sampling frame? institution community other not reported |
| 33. If a survey, what was the response rate? 80-100% 60-79% less than 50% not reported |
| 34. Indicate the prevalence |
| Abbreviation/Acronym | Definition |
|---|---|
| 2PDTs | 2-Point Discrimination Thresholds |
| ADS | Acceptance of Disability Scale (ADS) |
| AE | Adverse Effects |
| AHRQ | Agency for Healthcare Research and Quality |
| ASIA | American Spinal Injury Association Impairment Scale |
| BDI | Beck Depression Inventory |
| CA DREZ | Computer Assisted, Radiofrequency Dorsal Root Entry Zone |
| CES-D | Center for Epidemiological Study-Depression Scale |
| CHART | Craig Handicap Assessment and Reporting Technique |
| CNP | Central Neuropathic Pain |
| CNS | Central Nervous System |
| CO2 | Carbon Dioxide |
| CONSORT | Consolidation of the Standards of Reporting Trials |
| COPE | Coping Strategy Scales - Carver et al., 1989 |
| CSCM | Consortium for Spinal Cord Medicine |
| d | days |
| DBS | Deep Brain Stimulation |
| DCS | Dorsal Column Stimulation |
| DPS | Dysesthetic Pain Syndrome |
| Dx | diagnosis |
| DREZ | Dorsal Root Entry Zone |
| EMG | Electromyographic |
| EPC | Evidence-based Practice Center |
| FIM | Functional Independence Measure |
| GSW | gunshot wound |
| HPT | Heat Pain Threshold |
| LSIA | Life Satisfaction Index - A |
| m | months |
| MMPI | Minnesota Multiphasic Personality Inventory |
| MPI | Multidimensional pain inventory |
| MPQ | McGill Pain Questionnaire |
| MU-EPC | McMaster University Evidence-based Practice Center |
| MQS | Medication Quantification Scale |
| MVA | Motor Vehicle Accident |
| Nd: YAG | Neodymium: yttrium aluminum garnet |
| NMDA | N-methyl-D-aspartate |
| NS | Not Statistically Significant |
| NSAIDs | Non-Steroidal Anti Inflammatory Drugs |
| NR | Not Reported |
| NRS | Numerical Rating Scale |
| NWC | McGill subscale: Number of Words Chosen |
| PAD | Zung Pain and Distress Scale |
| PAG | periaqueducted gray stimulation |
| PES | Pain Experience Scale |
| PET | Positron Emission Tomography |
| PIS | Pain Inventory Scale |
| PPI | McGill subscale: Present Pain Intensity |
| POMS | Profile of Mood States |
| PRI | McGill subscale: Pain Rating Index |
| PSCS | Percutaneous insertion of electrodes into epidural space |
| PVA | Paralyzed Veterans Association |
| PVG | periventricular grey stimulation |
| pts | patients |
| QOL | Quality of Life |
| Rx | treatment |
| rCBF | Regional Cerebral Blood Flow |
| RCT | Randomized Controlled Trial |
| RSDS | Reflex Sympathetic Dystrophy Syndrome |
| SAPB | Subarachnoid phenol block |
| SCI | Spinal Cord Injury |
| SCIIS | Spinal Cord Injury Interference Scale |
| SCS | Spinal Cord Stimulation |
| SD | Standard Deviation |
| SE | Standard Error |
| SHS | Shoulder Hand Syndrome |
| SPECT | Single Emission Computed Tomography |
| SPI | Sternbach Pain Intensity |
| SSR | Sympathetic Skin Response |
| SSRIs | Selective Seratonin Reuptake Inhibitors |
| STAI | State-Trait Anxiety Inventory |
| TENS | Transcutaneous Electrical Nerve Stimulation |
| TOO | Task Order Officer |
| TSCI | Traumatic Spinal Cord Injury |
| UTI | Urinary Tract Infection |
| VAS | Visual Analogue Scale |
| w | weeks |
| y | years |
Free Full text in PMC].Free Full text in PMC]Free Full text in PMC]
[PubMed].Free Full text in PMC]Free Full text in PMC]Free Full text in PMC].Free Full text in PMC]Free Full text in PMC]Free Full text in PMC].Free Full text in PMC].Free Full text in PMC]Free Full text in PMC].Free Full text in PMC]Free Full text in PMC]Free Full text in PMC]
[PubMed].Free Full text in PMC]Free Full text in PMC]
[PubMed].Free Full text in PMC]Free Full text in PMC]
[PubMed].TELEform Standard Version 6.0, Cardiff Software Incorporated, San Marcos CA, USA.
