Effectiveness of spinal cord stimulation for the management of neuropathic pain

Transcription

1 Effectiveness of spinal cord stimulation for the management of neuropathic pain August 2009 Author: Peter Larking PhD

2 Important Note This evidence-based review summarises information on spinal cord stimulation for the management of neuropathic pain. It is not intended to replace clinical judgement, or be used as a clinical protocol. A reasonable attempt has been made to find and review papers relevant to the focus of this report. It does not claim to be exhaustive. This document has been prepared by staff of the ACC, Evidence Based Healthcare Advisory Group. The content does not necessarily represent the official view of ACC or represent ACC policy. ii

3 Executive Summary Background: The effectiveness of spinal cord stimulation (SCS) is reviewed for the clinical management of neuropathic pain related to Failed Back Surgery Syndrome (FBSS) and Complex Regional Pain Syndrome (CRPS-1). In each case SCS is compared to conventional medical management (CMM), reoperation or physical therapy (PT). Pain of ischaemic origin is excluded. The review is based substantially on the findings of a Health Technology Assessment report 1 conducted on behalf of the National Institute for Health and Clinical Excellence (NICE) 2. SCS requires the positioning of an lead/electrode in the epidural space at the level of the nerve roots innervating the painful area. A pulse generator (battery) connected to the lead supplies a current that induces a sensation that suppresses the pain. Search strategy: The basis of the search was for scientific literature published since the date of the HTA search (Sept 2007) and cited in Medline, DARE, CDSR, CCTR, EMBASE, and other evidenced based healthcare databases. The principal focus of these searches were for systematic reviews and randomised controlled trials (RCTs) not already covered in the NICE report (Appendix 1). Selection criteria: Types of Studies: The NICE and HTA reports, and systematic reviews, EBH reviews and/or guidelines and randomised controlled studies from Sept 2007 (the date of the HTA literature search). Types of Participants/Patients : Patients with neuropathic pain related to FBSS or CRPS-1. Types of Interventions: Spinal cord stimulation Types of Outcome Measures: Measures of pain relief and other clinical and functional outcomes. Main results: One systematic review and related guideline published in 2009 from the American Pain Society 3 and an additional paper reporting outcomes for a relevant RCT at two years 4 were found that were not covered by the HTA/NICE reports. The American Pain Society review was confined to FBSS only. NICE found evidence from the three RCTs to conclude that SCS was more effective than reoperation or CMM in reducing the chronic neuropathic pain of FBSS and more effective compared to physical therapy for CRPS 1. NICE concluded that SCS was a cost effective use of healthcare resources. The American Pain Society reviewing the same two RCTS for FBSS recommended that clinicians discuss risks and benefits of spinal cord stimulation as an option for treatment. iii

4 Level of evidence for effectiveness compared to reoperation + CMM or CMM was assessed as fair and net benefit, moderate. The two RCTS reviewed by HTA/NICE and the American Pain Society investigating SCS for FBSS compared the effectiveness of SCS for radicular neuropathic pain to reoperation (the North trial 5-8 ) and in the other to CMM (the PROCESS trial ). In the North trial, fifty patients with FBSS were randomised to either SCS + CMM or reoperation + CMM. The reoperation group were more likely to have crossed over at six months to SCS (14 of 26, 54%) compared to the SCS group crossover to reoperation (5 of 24, 21%) (p=.02). 50% pain relief and patient satisfaction was achieved by significantly more patients is the SCS group than in the reoperation. At an average of 2.9 years post operation, 45 were available for follow up; SCS was more successful than reoperation measured by a 50% reduction in VAS score; 47% of SCS achieved this goal versus 11.5% in the reoperation group (P<.01). In the PROCESS trial one hundred patients with FBSS were randomised to receive SCS + CMM or CMM. At 6 months, 48% in the SCS group and 9% in the CMM group had greater than 50% pain relief (P <.001). On a post hoc intention to treat (ITT) basis at one year 34% of the SCS group achieved a greater than 50% reduction in pain compared to 7% in CMM (P<.005). At 24 months this goal was achieved in 37% of patients randomised to SCS compared to 2% for CMM (p<.003). In the Kemler trial SCS/physical therapy was compared to physical therapy in an RCT involving fifty-four patients with CRPS-1. VAS mean score on an intention to treat basis was reduced at six months by 2.4 (scale 0 10) in the treatment group and rose 0.2 in the controls (P<.001). A similar result was found at two years but there was no difference between the groups at five years. There was no difference between the groups with respect to function at six months, two years or five years. Implantation of an SCS system is relatively safe and reversible. Costs are associated with complications which occur relatively frequently, such as lead migration and breakages. Overtime the effectiveness of SCS in many individuals appears to decline. The HTA reported 1 a cost/benefit analysis and concluded for FBSS that SCS compared to CMM or reoperation is a cost-effective intervention that would cost less than 20,000 per quality of life adjusted year. The cost effectiveness of SCS for CRPS-1 was less attractive and in some cases above 30,000 per QALY. The NICE Committee 2 in reviewing the HTA report concluded that SCS for treatment of FBSS and CRPS-1 would be a cost effective use of NHS resources. Conclusions: On balance it is concluded that in carefully selected patients with persistent and disabling radicular pain following surgery (FBSS) who have not responded to CMM after a period of at least six months, and in patients with CRPS-1 who have not responded to CMM including physical therapy after a period of at least six months that SCS will provide greater than 50% pain relief in more patients than would be obtained with continued standard treatment. iv

5 The evidence suggests that this balance in favourable outcomes can only be assured over intermediate time frames (three years) and in the case of CRPS-1 may not apply at 5 years. It would seem reasonable to conclude that ACC should continue to purchase SCS on a case by case basis for treating neuropathic pain related to FBSS and CRPS-1, but that monitoring of outcomes and costs to establish evidence of effectiveness in the New Zealand setting should be made a contractual obligation. Table of Contents 1 Background Research Questions Background and Health Technology Methodology Results Discussion Conclusions Appendix 1: Spinal cord stimulation: search strategies Appendix 2: Level of evidence in the SIGN system Appendix 3: Characteristics of included studies Appendix 4: Costs of typical systems References 28 Abbreviations SCS Spinal Cord Stimulation FBSS Failed Back Surgery Syndrome PT physical therapy CRPS-1 Complex Regional Pain Syndrome Type 1 CMM conventional medical management NICE National Institute for Health and Clinical Excellence HTA Health Technology Assessment, part of the National Institute for Health Research England ITT Intention to treat v

6 1 Background This study reviews the effectiveness of spinal cord stimulation (SCS) for the clinical management of neuropathic pain related to Failed Back Surgery Syndrome (FBSS) and Complex Regional Pain Syndrome (CRPS). SCS was compared to CMM, reoperation or PT. Pain of ischaemic origin is excluded. The review is based substantially on the findings of a Health Technology report 1 conducted on behalf of the National Institute for Health and Clinical Excellence (NICE) 2. 2 Research Questions What is the clinical effectiveness of spinal cord stimulation in the management of chronic neuropathic, radicular pain for FBSS, and for CRPS 1? What is the cost effectiveness of this treatment? 3 Background and Health Technology Spinal cord stimulation SCS requires the positioning of a lead in the epidural space at the level of the nerve roots innervating the painful area. A pulse generator (battery) supplies a current to a lead/electrode that induces a sensation that suppresses the pain (parathesia). The electrode impulses are thought to activate pain-inhibiting neuronal circuits within the dorsal horn and induce a perceived tingling sensation suppressing the pain 15. A basic tenet of SCS is to create an overlapping of the paresthesia and the pain region 16. At a chemical level SCS triggers the release of serotonin, substance P and GABA within the dorsal horn 16. A typical SCS system is composed of an electric pulse generator surgically implanted under the skin (often the abdomen), leads from the generator to lead/electrodes planted near the spinal cord and a hand held remote controller used by the patient to control the generator. The lead/electrode may be placed percutaneously within the epidural space or via a surgical laminectomy. SCS is a moderately invasive but readily reversible treatment, used only after more conservative treatments have failed. Usually, placement of a stimulator follows a trial stimulation to assess the ability of the device to cover the patient s area of pain with the parathesia. These tests do not guarantee that the effect will be maintained after the test nor is there evidence that such testing successfully predicts outcome long-term. It does identify patients though who don t like the sensation or cannot achieve appropriate stimulation. About 17 20% of patients fail the trial stimulation. Other prerequisites required may include, of adult age, failure of about 6 months of conventional management for pain, remedial surgery not feasible or advisable, no major psychiatric disorder and a willingness to eliminate any inappropriate drug use prior to implantation 17. ACC has prepared two previous reports on SCS in 2002 and as part of the Interventional Pain Management Guideline in The current purchasing recommendation is that SCS should be purchased on a case by case basis in those centres (Auckland and Christchurch) where protocols are currently in place and which collect procedural data that is available to guide ACC in future purchasing decisions. 6

7 Definitions: The NICE report 2 uses the following definitions of pain taken from the International Association of Pain (IASP) as follows - - Neuropathic pain is defined as pain initiated or caused by a primary lesion or dysfunction in the peripheral or central nervous systems. - FBSS is clinically defined as persistent or recurrent pain, mainly in the lower back and legs, after technically and anatomically successful lumbosacral spine surgery.it is sometimes referred to as persistent pain following (technically satisfactory) surgery. FBSS has both neuropathic and nociceptive components of pain. Nociceptive pain is caused by an injury to body tissues, and is outside the scope of this report. - Complex Regional Pain Syndrome is divided into two types. IASP has defined CRPS type I as usually following an initiating noxious event or period of immobilisation and satisfying the criteria of: continuing pain, allodynia (lowered pain threshold) or hyperalgesia (increased pain response); and oedema (accumulation of tissue fluid), changes in skin blood flow, or abnormal sudomotor activity (nerves that stimulate sweat glands) in region of the pain; and no existing conditions that would otherwise account for the degree of pain and dysfunction. CRPS type II follows nerve injury. IASP defines it as satisfying the three criteria; continuing pain, allodynia, or hyperalgesia after nerve injury, usually but not necessarily limited to the distribution of the injured nerve; and oedema, changes in skin blood flow, or abnormal sudomotor activity in region of pain; and no existing conditions that would otherwise account for the degree of pain and dysfunction. 4 Methodology Criteria for selecting studies for this review Types of Studies: The NICE report 2. EBH systematic reviews and/or guidelines, randomised controlled studies from Types of Participants/Patients : Adult patients with neuropathic pain, non ischaemic leg pain with FBSS or CRPS-1 who have not responded adequately to CMM/PT or surgical management. Types of Interventions: Spinal cord stimulation Types of Outcome Measures: Measures of pain relief and other clinical and functional outcomes. Criteria for excluding studies Studies other than RCTs were excluded and reports not written in English. Search Strategy and information sources This report is based on the NICE report 2 and literature published from and cited on the Medline, DARE, CDSR, CCTR, EMBASE, and other Evidenced based healthcare websites. (The NICE report was based on literature searches to Sept 07). 7

8 The principal focus of these searches was for RCTs not already covered in the NICE report. (See Appendix 1) Methods of the review All studies found were independently assessed by analysing the information including the definition of study population, description of intervention, criteria of inclusion and exclusion, methods of randomisation, the use of control group, baseline measurements of the study subjects in different study groups, blinding, sample size, outcome measurements, period of follow-up, compliance to the intervention, and potential bias in the studies. Papers were then graded using the SIGN convention ( and Appendix 3). This review is based mainly of the findings of the HTA report 1 and NICE recommendations 2. A separate literature search was conducted to find any additional systematic reviews or RCTs published since September 2007 the date of the last HTA search. Only one further publication was found which reported on 24 month follow up to the PROCESS trial 4 (see below) and a systematic review from the American Pain Society 3. 5 Results Description of studies Systematic Reviews 1. The NICE Report (2008) 2 was based on a Health Technology Assessment (HTA) 1 conducted at the University of Sheffield and reviewed papers published to September The Assessment evaluated SCS for chronic pain of neuropathic or ischaemic origin. The clinical questions covered were well defined and the search methodology was reported and comprehensive. Selection of relevant papers was well founded with appropriate inclusion and exclusion criteria. Methodology of the review was described and comprehensive. Conclusions drawn appear appropriate to the evidence found. The researchers found and reviewed three RCTs, for neuropathic pain; two relating to SCS for FBSS and one for CRPS-1, conducted in European, Australian and Canadian settings. All three studies used Medtronic implantable pulse generators and had test trials. The characteristics of the three prospective RCTs (North 5-8, PROCESS and Kemler ) which form the basis for HTA 1 an American Pain Society 3 reports are set out below and in Table 1. NICE, basing their recommendations on the HTA report found evidence from the three RCTs reviewed to conclude that SCS was more effective than reoperation or CMM in reducing the chronic neuropathic pain of FBSS and more effective than PT for CRPS 1. The cost effectiveness of SCS for FBSS and CRPS-1 is reviewed below. It is concluded that the HTA systematic review was of high quality and its conclusions appropriate for the settings in which the trials were held (SIGN score 1+). 8

9 2. The American Pain Society has reviewed the evidence base for interventional therapies, surgery and interdisciplinary rehabilitation for low back pain that included SCS. The review was developed into a Clinical Practice Guideline for the American Pain Society. The evidence was based on systematic reviews developed by the Oregon Evidence Based Practice Centre The clinical questions covered were well defined and the search methodology was reported and comprehensive. Selection of relevant papers was well founded with appropriate inclusion and exclusion criteria. Methodology of the review was described and comprehensive. Conclusions drawn appear appropriate to the evidence found. The review found only two RCTs related to radicular pain for FBSS, the North 5-7 and PROCESS trials, as did HTA. The clinical panel from the American Pain Society based on the systematic reviews made the following conclusion 3 : In patients with persistent and disabling radicular pain following surgery for herniated disc and no evidence of a persistently compressed nerve root, it is recommended that clinicians discuss risks and benefits of spinal cord stimulations as an option. Level of evidence for effectiveness compared to reoperation or CMM was assessed as fair and net benefit moderate. (Fair meant evidence was sufficient to determine effects on health outcomes but the strength of the evidence is limited by the number, quality and size of the studies). It was recommended that shared decision-making regarding SCS include discussion about the high rate of complications following SCS placement, cost, potential benefits and risk of complications. (There was insufficient evidence for the effectiveness of SCS for other types of failed back surgery syndrome or for low back pain with or without leg pain). This was a well conducted systematic review that has drawn a very cautious recommendation (1+). Randomised Controlled Trials 3. The North trial 5-8 was funded by the manufacturer. Fifty adults selected for the trial had received surgery for nerve root compression and had concordant complaints of persistent or recurrent radicular pain with or without low back pain meeting criteria for reoperation. Method of randomisation was reported but method of allocation and concealment was not. Patients were randomised to either SCS + CMM or reoperation + CMM. The trial was not blinded. Those failing the test screen were excluded from the trial (29%). The analysis was not based on an intention to treat methodology. The primary outcome was crossover to the other treatment option. Patients randomised to the the reoperation group were more likely at six months to have crossed over to SCS (14 of 26, 53.8%) compared to those in the SCS group crossing over to reoperation (5 of 24, 20.8%) (p=.02). 50% pain relief and patient satisfaction was achieved by significantly more patients is the SCS group than in the reoperation group. At an average of 2.9 years post operation 45 were available for follow up; SCS was more successful than reoperation as measured by a 50% reduction in VAS score; 47% 9

10 in the SCS (9 of 19) versus 11.5% in the reoperation group (3 of 26) P<.01. The reoperation group required an increase in opiate analgesics significantly more often than the SCS group (p<.025). There was no difference in employment status between groups. Nine percent of permanent implants required revision due to technical problems. The quality of the study was modest in that numbers were relatively small, allocation and concealment not disclosed and analysis was not based on the intention to treat principle (SIGN score 1-). 4. The PROCESS trial was funded by the manufacturer who also supplied logistical support, conducted the analysis and provided two of the six members to the trial committee. The key paper was written by independent members. The trial was conducted in 12 centres including Australia, United Kingdom, Europe and Canada. They recruited 100 adult patients with neuropathic pain of radicular origin in the legs. Pain intensity had to be greater than 50mm on the VAS scale. All patients had pain duration of at least 6 months after surgery for herniated disc refractory to CMM. Method of randomisation was reported and allocation and concealment appeared adequate. The trial was not blinded. Patients were randomised to either CMM or SCS + CMM. The trial at 6 months post implantation was analysed on an intention to treat basis and follow up for those receiving SCS was extended to 24 months (results not incorporated in HTA report 4 ). The primary outcome was a reduction of pain by more than 50% at 6 months as measured by the VAS scale. Crossover after 6 months was allowed. The key results were - Five of the nine participants failing the screen test actually went on to receive the SCS implant 11 on the patient s request. At 6 months 48% in the SCS group had greater than 50% pain relief compared to 9% in the CMM group (P <.001). VAS pain scores in the leg at 6 months were 39.9 in the SCS group compared to 66.6 in the CMM group (p<.0001); the corresponding results for back pain were 40.6 for SCS and 51.6 for CMM (p<.008). SF 36 results were significantly better in the SCS group for 7 of the 8 domains of the score. The domain that was not different was Role physical. Patient satisfaction with pain relief was 66% for SCS and 18% for CMM (p <.001). There was no significant difference in drug use. By 12 months 5 of 50 SCS patients (10%) crossed to CMM and 32 of 48 CMM patients (66.6%) crossed to SCS. On a per treatment basis at one year, 48% of the 71 patients implanted with a stimulator and 18% of the17 on CMM had >50% pain relief (P <.03). On a post hoc ITT basis 34% of the SCS group achieved a greater than 50% reduction in pain compared to 7% in CMM (P<.005). At 24 months this goal was achieved in 37% of patients randomised to SCS compared to 2% to CMM (p<.003). Over 24 months the 84 patients who received SCS, 27 (32%) had a total of 40 device complications. For 20 patients, surgery was required to resolve the event. Principal events were electrode migration (10%), infection or wound breakdown (8%) or a loss of paresthesia (7%). 35% of the SCS group and 52% of the CMM group 10

11 experienced a adverse drug event or development of a new illness, injury or condition. This RCT appears to have been well designed and conducted (SIGN score 1+). 5. In the Kemler study adult patients with CRPS 1 with mean pain intensity >5 (scale 0 10) and a duration greater than 6 months without a sustained response to CMM/PT were recruited into an RCT to receive SCS plus PT or PT. Method of randomisation was reported and allocation and concealment appeared adequate. Thirty six entered the SCS arm and 18 the PT arm. 33% failed the SCS screening. The trial was not blinded. At six months post implantation pain as measured by VAS was significantly more reduced by SCS than physical therapy alone (p=.001) with a similar result at two years (p=.001). At six months VAS mean score on an intention to treat basis was reduced by 2.4 (scale 0 10) in the treatment group and rose 0.2 in the controls (P<.001). 18/36 in the SCS group had a reduction in VAS score of greater than 50%. No difference in pain score was found between the two groups at five years. In contrast there was no difference between the groups with respect to function at six months, two years or five years. Over five years 10 of 24 patients underwent reoperation as a result of 29 complications 14. The 36 patients receiving SCS needed 42 pulse generators indicating a mean battery life of four years. The RCT was well designed and conducted (SIGN score 1+). Trial name Table 1. Characteristics of the three RCTS reviewed by NICE for neuropathic pain. Diagnosis Intervention /Control North FBSS SCS /reoperation Number entering trial Follow up North Mean 2.9 years PROCESS FBSS SCS + CMM + CMM Number remaining SCS/Control Outcome SCS/Control 60 6 months 23/26 Crossover 20.8%/80% P< /26 (Reduction in pain score of 50% or more) 47%/11.5% P< months 50/44 (Reduction in VAS of 50% or more) 48%/9% P=

12 PROCESS 12 months 47/41 34%/7% P=0.005 PROCESS 24months 46/41 37%/2% Kemler CRPS - 1 SCS + Physical Therapy /Physical Therapy P= months 36/18 (Reduction in VAS pain score on a scale of 10cm) 2.4 cm/+ 0.2 p<0.001 Kemler 2 years 35/ cm/0 P=0.001 Kemler 5 years 31/ cm/1.0 P=0.25 Cost Effectiveness Cost effectiveness of SCS was investigated by HTA 1 for NICE 2. HTA conducted a systematic identification and evaluation of published studies as well as an independent evaluation. Studies were sought from 1996 to September 2007 and search strategy and inclusion and exclusion criteria appeared to be appropriate. Only one published United Kingdom study meet the criteria 20. This study evaluated the cost effectiveness of SCS compared to conventional, nonsurgical treatment for patients with FBSS using a European Healthcare perspective. Costs used in the model were derived from Canada. NICE also considered the results of an analysis submitted by the Association of British Healthcare Industries on behalf of three manufacturers. The HTA considered this analysis to have an adequate evidence base and to be appropriately conducted. A third independent analysis was prepared by the School of Health and Related Research at the University of Sheffield. As in the previous two reports a 15 year time horizon was used. Methodology was based on that of Taylor 20 with scenarios similar to that of the Association of British Healthcare Industries based on the data from the three RCTS considered in the NICE report. Using a 15-year horizon, 4 year device longevity, a device price of 7745 and a threshold of 20,000 per QALY the HTA study found: FBSS: SCS+CMM vs CMM The results of the analysis suggest that SCS+CMM compared to CMM alone produce more QALYs. When using a threshold of 20,000 per QALY the probability of SCS+CMM being cost effective is around 99%. FBSS: SCS+CMM vs re-operation+cmm 12

13 The results suggest that SCS+CMM compared to re-operation produce more QALYs. When using a threshold of 20,000 per QALY the probability of SCS+CMM being cost effective is 100%. CRPS: SCS+PT vs PT alone. The results suggest that the probability of SCS+CMM being cost effective at a 20,000 per QALY threshold is around 78%. NICE concluded that the cost analysis suggested that SCS for FBSS compared to CMM or reoperation is a cost-effective intervention and less than 20,000 per QALY. The cost of effectiveness of SCS for CRPS -1 is less convincing and in some cases above 30,000 per QALY. Furthermore the CMM used in the Kemler trial was different to that used by the NHS and further research was called for to establish clinical effectiveness. Safety SCS complications include electrode migration, lead fracture, loss of paraesthesia, dural puncture and infection. The NICE report claims that across 403 implanted patients (including those for ischaemia) in 10 trials the percentage of patients requiring resurgery to resolve complications ranged across the trials from 0% to 38%. Studies indicate that technical complications occur in 27.2% of cases, biological complications in 4%, and other complications in 5%. 82% of the technical complications were due to issues with leads. One quarter to one third of patients require revision operations 15. Current Hardware Costs Medtronic and ANS have provided costs for typical systems. On the basis that one claimant had implanted a rechargeable neurostimulator with percutaneous lead and intraoperative accessories, the hardware cost of one system would range from around $21,000 to $35,000 (Appendix 4). Hospitalisation, health personnel costs, costs of complications and replacements would be additional. Historic Utilisation, Costs and Outcomes At the date of writing of this report no data on past utilisation, costs or outcomes was available. 6 Discussion Methodological Quality Study design: The HTA/NICE 1 2 and American Pain Society 3 reviews appear to have been thorough, unbiased in their approach and well conducted. The literature searches were comprehensive and appear to have included all relevant RCTs in English to date. The conclusions to the reviews are reasonable in terms of the evidence available. The three 13

14 cost analyses reviewed in the HTA report used standard methodologies and appear to be appropriate. The three RCTs on which the HTA report was based were of reasonable quality. The PROCESS and Kemler trials were adequately powered, randomisation and allocation concealment were described, the comparison group appropriate, followup ranged from 2 to 5 years and outcome measures appropriate to the research questions were asked. Both trials analysed data on an intention to treat basis. The North trial 5-8 was of poorer quality in that allocation concealment was not described, and analysis was not based on the intention to treat basis. Interpretation of the longer term results is complicated by the high crossover rate at 6 months when 54% of those receiving reoperation crossed over to receive SCS and 21% of the SCS group crossed over to re-operation 21. Total numbers treated in the trials were not large and the trials considered are heterogeneous especially with respect to the interventions the control groups received. All three RCTs have been criticised for having as the control group a treatment option (reoperation, CMM or physical therapy) which patients had already received and experienced failure 21. (Oddly enough despite the past experience of failure small but significant numbers in these control groups then go on in the course of the trial to experience greater than 50% pain relief). Further criticism points to the use of what CMM means, in that it is not a single therapy and is not likely to be of consistent or uniform practice 22. It has also been noted that no RCT has compared effectiveness of SCS with intensive interdisciplinary rehabilitation 3. Clinical Effectiveness It would appear that many, carefully selected patients with neuropathic pain related to FBSS or CRPS-1 will benefit from reduced pain over two year to three years post implantation compared to standard treatments but longer term benefit is less certain. Cost effectiveness The conclusion of NICE that SCS for FBSS and CRPS-1 is a cost effective use of NHS resources was obviously a balanced judgement based on reasonable evidence. Costs however were identified as being influenced by a number of factors such as the probability of achieving optimal pain relief with SCS, the duration of benefit, device longevity and device cost. Little appears to be known about these factors in the New Zealand situation and therefore there is no certainty that the NICE conclusions are applicable here. Safety and adverse effects SCS though somewhat invasive is a reversible procedure. The main complications such as infection and hardware complications are all manageable but do affect the cost effectiveness. Limitations of the review Only RCTs in English were included. The main limitations of the review relate to the settings in which the interventions were delivered; these may not reflect the New Zealand scene especially in relation to cost inputs, policy and practice with respect to reoperation for FBSS, and practice of conventional medical management. 14

15 7 Conclusions The conclusions of NICE and American Pain Society were as follows: NICE 2 concluded that SCS was more effective than reoperation or CMM in reducing the chronic neuropathic pain of FBSS, and CRPS 1 more effective than PT. They also found that the cost analysis suggested that SCS for FBSS compared to CMM or reoperation was a cost-effective intervention and less than 20,000 per QALY. The cost of effectiveness of SCS for CRPS-1 was less convincing and in some cases above 30,000 per QALY. The American Pain Society 3 concluded that for patients with persistent and disabling radicular pain following surgery for herniated disc and no evidence of a persistently compressed nerve root, clinicians discuss risks and benefits of spinal cord stimulations as an option. Level of evidence for effectiveness compared to reoperation or CMM was assessed as fair and net benefit moderate. On balance it is concluded that in carefully selected patients with persistent and disabling radicular pain following surgery who have not responded to CMM or reoperation after a period of at least six months, and in patients with CRPS-1 who have not responded to CMM including physical therapy, that SCS will provide greater than 50% pain relief pain relief in more patients than would be obtained with continued standard treatment. The evidence suggests that this balance in favourable outcomes can only be assured over intermediate time frames (three years) and in the case of CRPS-1 may not apply at 5 years. In the United Kingdom setting NICE, concluded that for the above patients, SCS was a cost effective use of health resources. However cost effectiveness cannot necessarily be inferred in the NZ setting. The strength of evidence for the above statements is only of a moderate level. It is clear that both NICE and the American Pain Society have both carefully weighed the evidence and considered SCS to be an effective treatment in for the patients under consideration. There is little risk of adverse outcomes or complications that are not manageable although their occurrence may affect cost effectiveness (There incidence in the NZ setting is not known). It would seem reasonable to conclude that ACC should continue to purchase on a case by case basis, SCS for treating neuropathic pain related to FBSS and CRPS-1 but that efforts should be made to monitor outcomes and collect costing data to establish evidence of effectiveness in the New Zealand setting. It is suggested that data collection and reporting be made a contractual obligation. 15

16 Appendix 1: Spinal cord stimulation: search strategies May 2009 by Emma Roache Medline 1. spinal cord stimulation.mp. 2. Electric Stimulation Therapy/ 3. spinal cord injuries/ or spinal cord/ 4. 1 or (2 and 3) 5. failed back surgery.mp. 6. chronic regional pain.mp. 7. neuropathic pain.ti,ab,sh. 8. exp complex regional pain syndromes/ or causalgia/ or reflex sympathetic dystrophy/ 9. back pain/ or failed back surgery syndrome/ or low back pain/ or 6 or 7 or 9 or and Randomized controlled trials/ 13. Randomized controlled trial.pt. 14. Random allocation/ 15. Double blind method/ 16. Single blind method/ 17. Clinical trial.pt. 18. exp clinical trials/ 19. or/ (clinic$ adj trial$1).tw. 21. ((singl$ or doubl$ or treb$ or tripl$) adj (blind$3 or mask$3)).tw. 22. Placebos/ 23. Placebo$.tw. 24. Randomly allocated.tw. 25. (allocated adj2 random).tw. 26. or/ or Case report.tw. 29. Letter.pt. 30. Historical article.pt. 31. Review of reported cases.pt. 32. Review, multicase.pt. 33. or/ not and limit 35 to (english language and yr="2007 -Current") EMBASE 1. spinal cord stimulation/ 2. failed back surgery syndrome/ 3. failed back surgery.ti,ab. 4. exp complex regional pain syndrome/ or exp complex regional pain syndrome type i/ or exp complex regional pain syndrome type ii/ 5. exp Backache/ 6. Neuropathic Pain/ 7. 6 or 4 or 3 or 2 or and 7 9. Clinical trial/ 10. Randomized controlled trial/ 11. Randomization/ 16

17 12. Single blind procedure/ 13. Double blind procedure/ 14. Crossover procedure/ 15. Placebo/ 16. Randomi?ed controlled trial$.tw. 17. Rct.tw. 18. Random allocation.tw. 19. Randomly allocated.tw. 20. Allocated randomly.tw. 21. (allocated adj2 random).tw. 22. Single blind$.tw. 23. Double blind$.tw. 24. ((treble or triple) adj blind$).tw. 25. Placebo$.tw. 26. Prospective study/ 27. or/ Case study/ 29. Case report.tw. 30. Abstract report/ or letter/ 31. or/ not and limit 34 to (english language and yr="2007 -Current") All EBM Reviews - Cochrane DSR, ACP Journal Club, DARE, CCTR, CMR, HTA, and NHSEED 1. spinal cord stimulation.ti,sh,kw,hw,oh,kf. 2. Electric Stimulation Therapy/ 3. back pain/ or spinal cord.ti,sh,kw,hw,oh,kf and or 1 6. failed back surgery syndrome.ti,sh,kw,hw. 7. complex regional pain syndrome.ti,sh,kw,hw,oh,kf. 8. causalgia.ti,sh,kw,hw,oh,kf. 9. reflex sympathetic dystrophy.ti,sh,kw,hw,oh,kf. 10. backache.ti,sh,kw,hw,oh,kf. 11. Back pain.ti,sh,kw,hw,oh,kf. 12. low back pain.ti,sh,kw,hw,oh,kf. 13. Neuropathic Pain.ti,sh,kw,hw,oh,kf or 11 or 7 or 9 or 12 or 8 or 10 or and limit 16 to yr="2007 -Current" CINAHL S22 S20 and S21 S21 Limiters - Publication Year from: S20 S10 and S19 S19 S12 or S13 or S14 or S15 or S16 or S17 or S18 TX trebl* mask* or TX trebl* blind S17 TX allocat* random* or SU quantitative studies or SU placebos S16 SU random assignment or TX random* allocat* or TX placebo* S15 TX singl* mask* or TX doubl* mask* or TX tripl* mask* S14 TX singl* blind* or TX doubl* blind* or TX tripl* blind* S13 PT clinical trial or TX clinic* w1 trial* or TX randomi?ed control* trial* 17

18 S12 (MH "Clinical Trials+") S11 S9 and S10 S10 S3 or S4 S9 S5 or S6 or S7 or S8 S8 "neuropathic pain" S7 (MH "Back Pain+") S6 (MH "Complex Regional Pain Syndromes+") S5 "failed back surgery" S4 TI spinal cord stimulation or AB spinal cord stimulation S3 S1 and S2 S2 (MH "Electric Stimulation") or (MH "Transcutaneous Electric Nerve Stimulation") or (MH "Electrical Stimulation, Functional") or (MH "Electrical Stimulation, Neuromuscular") or (MH "Transcutaneous Electrical Nerve Stimulation (Iowa NIC)") S1 (MH "Spinal Cord") or (MH "Spinal Cord Injuries") Appendix 2: Levels of evidence in the SIGN system Table 1. SIGN criteria for classifying studies. Score Design 1++ High quality meta-analyses, systematic review of RCTs, or RCTs with a very low risk of bias 1+ Well conducted meta-analyses, systematic reviews of RCTs, or RCTs with a low risk of bias 1- Meta-analyses, systematic reviews of RCTs or RCTs with a high risk of bias 2++ High quality systematic reviews of case-control or cohort studies. High quality case-control or cohort studies with a very low risk of confounding, bias or chance and a high probability that the relationship is causal 2+ Well conducted case control or cohort studies with a low risk of confounding, bias or chance and a moderate probability that the relationship is causal 2- Case control or cohort studies with a high risk of confounding, bias or chance and a significant risk that the relationship is not causal. 3 Non-analytic studies 4 Expert opinion 18

19 Appendix 3: Characteristics of Included Studies Study Design Study Characteristics Intervention Outcomes Level of Evidence A systematic review (HTA) of Spinal cord Outcome measures were based on the three RCTs. Reviews 3 prospective stimulators with RCTs and included pain, quality of life, physical 1+ RCTS relating to neuropathic implantable and functional abilities, medication use, pain (two FBSS, one CRPS) pulse generator complications and adverse effects. (Study also reviewed SCS for pain systems of ischaemic origin but this is not part of the ACC report). An economic analysis was also performed. Simpson EL et al; (HTA) Settings: RCTS reviewed were conducted in clinics in USA, European, Canadian and Australian settings Reviewed all relevant RCTs up to Sept 07 Inclusions - RCTS trialling spinal cord stimulators in adults with chronic neuropathic pain (FBSS, CRPS-1) who have had an inadequate response to pain from medical, physical or surgical treatment. Comparison Groups: conventional medical management, physical therapy and reoperation. Exclusions: Neurostimulation of other parts of the nervous system. Articles reviewed were limited to English Comparison groups were medical and/or surgical treatments excluding SCS All 3 RCTs included test stimulation Results. Stimulators used were only of Medtronic brand The researchers found and reviewed three RCTs, for neuropathic pain; two relating to SCS for FBSS and one for CRPS-1, conducted in European and Canadian settings. The characteristics of the three prospective RCTs (North 5 6, PROCESS and Kemler ) is set out below. Heterogeneity precluded a metaanalysis. The HTA found evidence from the three RCTs reviewed to conclude that SCS was more effective than reoperation or conventional medical management (CMM) in reducing the chronic neuropathic pain of FBSS and CRPS 1 in carefully selected patients. Trials lasted up to 5 years. The cost analysis suggests that SCS for FBSS compared to CMM or reoperation is a costeffective intervention and less than 20,000 per QALY. The cost of effectiveness of SCS for CRPS -1 is less convincing and in some cases above 30,000 per QALY. Furthermore the CMM used in the Kemler trial was different to that 19

20 Study Design Study Characteristics Intervention Outcomes Level of Evidence used by the NHS and further research was called for to establish clinical effectiveness. Quality: A well conducted systematic review and cost/benefit analysis. Chou R et al (American Pain Soc) A clinical practice guideline based on a systematic review of RCTs relating to low back pain which included SCS. Reviews 2 prospective RCTS (North and PROCESS) relating to neuropathic pain for FBSS (Study also reviewed other interventions for low back pain but this is not part of the ACC report). Settings: RCTS reviewed were conducted in European, Canadian and Australian settings Reviewed all relevant RCTs up to July 2008 Inclusions - RCTS trialling spinal cord stimulators in adults with persistent chronic pain following surgery for herniated disc and no evidence of a persistently compressed nerve root. Comparison Groups: Conventional medical management, physical therapy and reoperation. Exclusions: Spinal cord stimulators with implantable pulse generator systems Comparison groups were medical and/or surgical treatments excluding SCS The 2 RCTs included test stimulation Biases and weaknesses, Relies on three smaller RCTS for its conclusions. Settings and costings may not be appropriate to the NZ setting. Outcome measures were based on the two RCTs and included pain, quality of life, physical and functional abilities, medication use and complications and adverse effects. Results. Stimulators used were only of Medtronic brand The researchers found and reviewed two RCTs, for neuropathic pain; conducted in European and Canadian settings. The characteristics of the two prospective RCTS (North 5 6, PROCESS ) is set out below. Heterogeneity precluded a meta-analysis. The clinical panel from the American Pain Society based on the systematic review made the following conclusion: 1+ Articles reviewed were limited to English In patients with persistent and disabling radicular pain following surgery for herniated disc and no evidence of persistently compressed nerve root, it is recommended that clinicians discuss risks and benefits of spinal cord stimulations as an option. Level of evidence for effectiveness was assessed as fair and net benefit moderate. (Fair meant evidence was sufficient 20

21 Study Design Study Characteristics Intervention Outcomes Level of Evidence to determine effects on health outcomes but the strength of the evidence is limited by the number, quality and size of the studies) Quality: A well conducted systematic review North RB, et al 5-8 Design: Prospective randomised controlled trial Not analysed on intention to treat basis Randomisation described Setting: Johns Hopkins University Blinding: The six month assessment was conducted by a disinterested party but not blinded. Allocation and concealment not described Selection and Dropouts: 99 patients eligible of whom 60 consented and 50 randomised. No dropouts during the first 6 months (one died near the end of the 6 months) Follow up: 6 months and an average of 2.9 years post op. Participants: Selected patients with FBSS Mean age: 52 % males 48% Duration of pain: Not stated Inclusions: Surgically remediable nerve root compression for radicular pain, with or without low back pain, refractive to conservative care after one or more lumbosacral spine surgeries. Exclusions: Major neurological deficit, radiographically critical neural compression, drug dependency, major psychiatric comorbidity, a chief complaint of axial pain exceeding radicular pain etc. Intervention: 50 selected patients randomised to receive reoperation + CMM or SCS + CMM Biases and weaknesses, Relies on two smaller RCTS for its conclusions. Settings and costings may not be appropriate to the NZ setting. Outcome measure: Frequency of crossover at six months. Patients were offered crossover to the alternative treatment at six months. Results: At 6 months patients initially randomised to SCS were significantly less likely to have crossed over than were those randomised to reoperation (5 of 24 (20%) cf to 14 of 26, 53.8%)) p < (the 5 in the SCS group who crossed over failed the initial SCS screen). At an average of 2.9 years post op 45 were available for follow up; SCS was more successful than reoperation measured by a 50% reduction in VAS score; 47% of SCS (9 of 19) versus 11.5% of reoperation group (3 of 26) P<.01 Patients randomised to reoperation required increased opiates cf to SCS patients p <.025 Measures of work status and activities of daily living did not differ. Complications: 9% who received permanent implants underwent hardware revisions 1-21

22 Study Design Study Characteristics Intervention Outcomes Level of Evidence After 6 months crossover was offered to those for whom treatment was unsuccessful. At average of 2.9 years post operation 45 of 50 (90%) available for review, 19 in SCS group and 26 in reoperation group Numbers: 26 randomised to reoperation and 24 to SCS Quality: An RCT with limited follow up. Methodological quality was modest. Settings, clinical practice and costings may not be applicable to the NZ setting. PROCESS Study Kumar K, et al Prospective randomised controlled trial Conducted in 12 centres including Canada, UK, Europe and Australia Randomisation described ITT analysis Settings: University hospitals, pain clinics, public hospitals. No Blinding Allocation and concealment described Selection and dropouts: Primary outcome data available for 50/52 in SCS group and 44/48 in CMM group. At 24 months 46/52 were available in SCS group and 41 of 48 in CMM. Follow up 6 months, 1 and 2 years Adult patients with neuropathic pain of radicular origin in the legs. (FBSS). Mean age 52 (CMM) and 48.9 (SCS) % males 44% (CMM), 58% (SCS) Time since last surgery 4.6 years (CMM), 4.7 (SCS) Inclusions: Pain intensity had to be greater than 50mm on the VAS scale. All patients had a history of nerve injury and pain duration of at least 6 months after surgery for herniated disc Exclusions mostly non radicular back pain, another disabling chronic Interventions CMM or CMM and SCS. At six months patients were allowed to crossover. Baseline characteristics were comparable except for VAS back pain score 44.8% (CMM) cf 54.5% (SCS). Primary outcome was >50% reduction in VAS at 6 months. Secondary outcomes - Reduction of leg and back pain, QOL, treatment satisfaction drug therapy. Results 6 months: 50% pain relief - 9% CMM, 48% SCS P <.001 Leg pain VAS score in CMM group was 66.6 v in SCS, P <.0001 Back pain VAS score was 51.6 in CMM group v SCS, p <.008 SF 36, 7 of 8 domains significantly better in SCS group, p >.01 for each of the 7. The domain not different was Role physical Patient satisfaction with pain relief 18% CMM, 66% SCS p <.001 Drug therapy no significant difference for opioids, NSAIDS, antidepressants. Anticonvulsants were used less in SCS group, p<

23 Study Design Study Characteristics Intervention Outcomes Level of Evidence pain condition etc Results at 12 months: 5 SCS patients crossed to CMM and 28 CMM patients crossed to SCS. On a per treatment basis 48% of the 71 patients implanted with a stimulator and 18% of the 17 on CMM had >50% pain relief P <.03. On a post hoc ITT basis at one year 34% of the SCS group achieved the primary outcome compared to 7% in CMM P<.005. Numbers: 100 of 214 selected 48 received CMM, 52 SCS. 9 in the SCS group failed the screen but five elected to have the device implanted. At 24 months; On a post hoc basis the primary outcome was achieved in 37% of SCS cf 2% CMM (p <.003) and 47% 0f 72 patients who received SCS as final treatment v. 7% of 15 for CMM (p=.02). Complications and adverse events at one year Of the 84 patients who received SCS 27 (32%) had a total of 40 device complications. For 20 surgery was required to resolve the event. Principal events were electrode migration (10%), infection or wound breakdown (8%) or a loss of paresthesia (7%). 35% of the SCS group and 52% of the CMM group experienced a adverse drug event or development of a new illness, injury or condition. Quality: A well conducted multicentred RCT. Methodological quality was good. Kemler MA, Barendse GAM, van Kleef K, devet HCW, et Design: Prospective, case control, randomised with stratification, and controlled. Analysed on an intention to treat basis Participants: Patients with CRPS 1., 26 precipitated by trauma, 24 by surgery and spontaneously in 4. Treatment Group: SCS + physical therapy Settings, clinical practice and costings may not be applicable to the NZ setting. Outcomes measured: Pain VAS and McGill Pain Questionnaire, Global perceived effect on 7 point scale Functional status Quality of life Nottingham Health Profile 1+ 23

24 al Study Design Study Characteristics Intervention Outcomes Level of Evidence Complications Setting: University Hospital Age: years Netherlands Blinding: For randomisation only. Allocation and concealment described Randomised in ratio of 2:1; SCS:PT Duration of pain: At least 6 months of pain Selection and Dropouts: 110 patients as potential candidates, Of the 36 assigned to SCS/Physical therapy, the trial screen was successful in 24 At 6 months no dropouts, At 2 years, 3 dropouts; one from SCS group in whom it was initially found that electrodes could not be placed and 2 from the physical therapy group who eventually received SCS. At 5 years 31/36 remained in SCS group (4 lost to follow up and 1 received a special implant) and 13/18 in PT (5 received implant and 1 lost to follow up) Exclusions: Raynaud s, other neurological abnormalities etc Substance abuse, major psychiatric illness, issues with possible secondary gain Numbers: 54 enrolled, 36 assigned to SCS/physical therapy group, 18 to physical therapy alone. Control Group: Physical therapy alone Results: At six months VAS mean score on intention to treat basis reduced by 2.4 in treatment group and rose 0.2 in controls (P<.001) 18/36 in SCS group had a reduction in VAS of > 50% of baseline Global perceived effect also significantly different between groups P<.01 Other changes in measures of pain, quality of life and functional status did not differ between groups At two years- Mean pain intensity reduced by 2.1 in SCS group cf to 0 in physical therapy group P <.001. At two years SCS was regarded as successful in 20 of 35 patients. Global perceived effect 43% SCS cf to 6% in PT p<.001 Quality of life and functional scores not different. At 5 years, there was no significant difference in pain relief between the groups nor with other variables measured. SCS was regarded as successful in 11 of 31 patients. Complications: Over 5 years, in SCS group 10 of the 24 patients (42%) underwent reoperation as a result of 29 complications. Follow up: Quality: 24