You are here

  • Home
  • Online First
  • Beyond traditional therapies: a network meta-analysis on the treatment efficacy for chronic phantom limb pain

Article Text

Article menu
Review
Beyond traditional therapies: a network meta-analysis on the treatment efficacy for chronic phantom limb pain
  1. Sun-Mei Chung1,2,
  2. Jui-Chien Wang2,3,
  3. Chung-Ren Lin4,
  4. Shu-Cheng Liu4,
  5. Po-Ting Wu5,6,
  6. Fa-Chuan Kuan5,7,
  7. Ching-Ju Fang8,9,
  8. Yu-Kang Tu10,11,
  9. Kai-Lan Hsu5,7,
  10. Pei-Chun Lai12 and
  11. Chien-An Shih5,6,7
  1. 1Mackay Memorial Hospital, Taipei, Taiwan
  2. 2School of Medicine, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  3. 3Taipei Veterans General Hospital, Taipei, Taiwan
  4. 4Department of Anesthesiology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  5. 5Department of Orthopedics, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  6. 6Department of Orthopedics, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  7. 7Medical Device R & D Core Laboratory, National Cheng Kung University Hospital, Tainan, Taiwan
  8. 8Department of Secretariat, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  9. 9Medical Library, National Cheng Kung University, Tainan, Taiwan
  10. 10Institute of Epidemiology and Preventive Medicine, National Taiwan University College of Public Health, Taipei, Taiwan
  11. 11Department of Dentistry, National Taiwan University Hospital, Taipei, Taiwan
  12. 12Education Center, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan, Taiwan
  1. Correspondence to Dr Chien-An Shih, Department of Orthopedics, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan 704302, Taiwan; sam5214330{at}hotmail.com

Abstract

Background Phantom limb pain (PLP) frequently affects individuals with limb amputations. When PLP evolves into its chronic phase, known as chronic PLP, traditional therapies often fall short in providing sufficient relief. The optimal intervention for chronic PLP remains unclear.

Objective The objectives of this network meta-analysis (NMA) were to examine the efficacy of different treatments on pain intensity for patients with chronic PLP.

Evidence review We searched Medline, EMBASE, Cochrane CENTRAL, Scopus, and CINAHL EBSCO, focusing on randomized controlled trials (RCTs) that evaluated interventions such as neuromodulation, neural block, pharmacological methods, and alternative treatments. An NMA was conducted based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. The primary outcome was pain score improvement, and the secondary outcomes were adverse events.

Findings The NMA, incorporating 12 RCTs, indicated that neuromodulation, specifically repetitive transcranial magnetic stimulation, provided the most substantial pain improvement when compared with placebo/sham groups (mean difference=−2.9 points, 95% CI=−4.62 to –1.18; quality of evidence (QoE): moderate). Pharmacological intervention using morphine was associated with a significant increase in adverse event rate (OR=6.04, 95% CI=2.26 to 16.12; QoE: low).

Conclusions The NMA suggests that neuromodulation using repetitive transcranial magnetic stimulation may be associated with significantly larger pain improvement for chronic PLP. However, the paucity of studies, varying patient characteristics across each trial, and absence of long-term results underscore the necessity for more comprehensive, large-scale RCTs.

PROSPERO registration number CRD42023455949.

  • Nerve Block
  • Pharmacology
  • analgesia
  • CHRONIC PAIN
http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, an indication of whether changes were made, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/rapm-2023-105104

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Phantom limb pain (PLP) is a common consequence of limb amputations, occurring in 60%–70% of cases.1 Of these individuals, 10%–15% experience severe pain episodes, while 50%–85% may develop chronic PLP.2 3 Among those with chronic PLP, up to 25% endure significant pain-related disability.4 As PLP advances to a chronic stage, treatment becomes more challenging due to persistent functional and structural alterations in pain pathways.5 Despite ongoing research, a definitive treatment for chronic PLP remains elusive, with fewer than 10% of patients achieving sustained relief from conventional treatments such as medications or epidural injections.6

    A wide range of treatments for chronic PLP exists,1–4 6–13 yet no standard treatment for chronic PLP has been established, making the most effective option remains challenging. These treatments encompass neuromodulation techniques such as repetitive transcranial magnetic stimulation (rTMS),11 cerebellar transcranial direct current stimulation (ctDCS),12 and peripheral nerve stimulation (PNS),13 established nerve-blocking methods such as continuous perineural block (CPNB)2 and cryoneurolysis,3 pharmaceutical options such as oral amitriptyline,9 gabapentin,4 memantine,1 mexiletine,10 and morphine,10 and other techniques, notably electromagnetic shielding (EMS).6 The absence of in-depth knowledge about the mechanisms of PLP presents challenges in establishing consistent clinical guidelines.14 Currently, only expert consensus guides the treatment of general PLP, emphasizing the importance of non-pharmacological treatments.

    Previous research, encompassing multiple systemic review and pairwise meta-analyses15–20 or a network meta-analysis (NMA),21 has evaluated treatments for PLP. However, these studies primarily focused on perioperative treatment and the general PLP,15–21 rather than honing in on the specificities of the “chronic” PLP subgroup. Addressing chronic PLP requires a more tailored therapeutic approach compared with standard PLP treatments.22 Moreover, although several randomized controlled trials (RCTs) have been established to gage the effectiveness of treatments for chronic PLP, a holistic multiarm comparative analysis has proven either intricate or clinically impractical. Consequently, this NMA aims to compare the clinical outcomes of different chronic PLP treatments, based on a systematic review and a detailed examination of recent RCT results.

    Methods

    Search strategy

    The NMA protocol was prospectively registered on PROSPERO (Registration number: CRD42023455949). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 extension guidelines for reporting the results of NMA in healthcare interventions. Our comprehensive database searches encompassed Medline, EMBASE, Cochrane CENTRAL, Scopus, and CINAHL EBSCO, spanning from inception to July 10, 2023, without language restrictions. In addition, we screened and incorporated references from relevant studies that met our inclusion criteria. Detailed search strategies are available in online supplemental appendix 2.

    Supplemental material

    Inclusion and exclusion criteria

    We incorporated all relevant RCTs assessing different treatment approaches for chronic PLP in individuals who have been experiencing pain for at least 2 months or more, or where the term “chronic PLP” was specifically mentioned. We excluded non-randomized trials, quasi-experimental designs, trials focused on preventive or immediate postoperative PLP treatments, single-arm trials, trials without predefined outcome measures, trials without accessible arm-level data, and trials with a duration of only a few minutes to hours.

    Data extraction and management

    Two authors (S-MC and J-CW) independently screened titles and abstracts of all entries that met our search criteria. Full texts were retrieved for selected trials to assess their eligibility for inclusion. Data extraction from the included RCTs was conducted using a predesigned data sheet, which captured the following information: authors’ names, publication year, journal of publication, study design, inclusion and exclusion criteria, intervention and control protocols, patient characteristics, outcome measures, and risk of bias. Any disagreements or conflicts between the authors were resolved through discussion or by seeking the judgment of the third author (C-AS).

    Type of intervention

    We considered interventions addressing chronic PLP and categorized them as follows: (1) neuromodulation, which comprises rTMS, ctDCS, and PNS; (2) nerve block, including CPNB and cryoneurolysis; (3) pharmacological treatments, such as oral amitriptyline, gabapentin, memantine, mexiletine, and morphine; and (4) alternative approaches, exemplified by EMS.

    Type of outcome measurement

    The primary outcome assessed was the change in pain intensity before and after treatment, which was measured using either the Numerical Rating Scale (NRS) or Visual Analog Scale (VAS). The secondary outcome focused on determining the total rate of adverse events for each individual intervention. Data were obtained from RCTs at the end of follow-up periods. For cross-over RCTs, data were extracted at the time point just before the cross-over occurred. However, in some trials that only presented pooled results for each intervention arm before and after cross-over, these pooled data were extracted.

    Addressing missing parameters

    In addressing missing parameters for this NMA, intention-to-treat analysis results were used. If mean values were missing for numerical variables, they were replaced with medians. SDs were derived from CIs when available, or else, IQRs were divided by 1.35 to estimate SDs. We also calculated the average values and SDs of the changes in pain scores when only baseline and follow-up measurements were available.23

    Quality assessment

    The Cochrane Collaboration’s RoB2 tool, comprising five domains and an overall risk assessment, was employed to assess bias risk.24 Two authors (SMC, JCW) independently reviewed and scored all included RCTs, categorizing them as “high risk,” “some concerns,” or “low risk” using RoB2. For cross-over RCTs, we applied the RoB2 framework for cross-over trials, which includes an additional domain, “Domain S: Bias arising from period and carryover effects.” In cases of disagreement, a third author (C-AS) provided input.

    Quality of evidence

    The Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for NMA was used to evaluate evidence certainty across five domains: study limitation, inconsistency/heterogeneity, indirectness, imprecision, and publication bias, assigning confidence ratings as high, moderate, low, or very low.25 26

    Publication bias

    For assessing publication bias, the presence of small-study effects was evaluated for each outcome using the comparison-adjusted funnel plot and Egger’s test.

    Data synthesis and statistical analysis

    Data synthesis and statistical analysis were conducted by using STATA V. 15.0 (StataCorp). A frequentist approach was employed for contrast-based model meta-analysis, integrating random-effects NMA to facilitate comparisons among multiple interventions, incorporating both direct and indirect evidence to enhance the robustness of estimates. The effect measures were reported as the mean difference (MD) with a 95% CI for changes in pain intensity, and as ORs with a 95% CI for adverse events. The ranking of interventions was determined using the surface under the cumulative ranking curve area (SUCRA).27 Inconsistency was assessed through various models, encompassing global inconsistency through design-by-treatment interaction models and local inconsistency through loop inconsistency models and node-splitting models.28 29 To validate the transitivity assumption, we scrutinized effect modifier distributions such as age, male percentage, and baseline VAS/NRS score. Heterogeneity was evaluated using I2 in pairwise meta-analysis, the tau value for between-study heterogeneity, and a comprehensive examination of study characteristics. We performed a meta-regression analysis to identify potential effect modifiers, drawing on thresholds established in previous studies concerning chronic pain and PLP.30 31 This process entailed categorizing data according to several criteria: baseline pain score (either above or below 5.8 points),30 patient age (either above or below 55 years),31 and duration postamputation (either more than or less than 2 years),31 and the predominant amputation site and type (accounting for more than 50%). Additionally, we conducted a sensitivity analysis by excluding trials that relied on imputed data, opting instead for those using the mean and SD to assess pain severity.

    Findings

    A total of 2975 studies were identified through database searches (figure 1). After removing duplicates and screening the titles and abstracts (online supplemental appendix 3), 12 studies1–4 6–13 were selected for inclusion in the analysis (table 1 and online supplemental appendix 4). Out of these, seven trials1 3 6 8 9 11 are RCTs, while the remaining five trials2 4 10 12 13 are cross-over RCTs. The assessment of transitivity is presented in online supplemental appendix 5. Regarding these trials, the risk of bias was evaluated as follows: two trials2 4 showed no concern, seven trials1 3 6 9 11–13 had some concerns, and three trials7 8 10 (online supplemental appendix 6). Among these trials, nine trials1–3 7–10 12 included patients with PLP lasting longer than 2 months, while the others four trials4 6 11 13 included patients with “chronic PLP” without stated chronic PLP duration. Information on adverse events was retrievable in eight trials.1 3 6 8 10–13 The duration since amputation was reported in eight trials.1 2 6–9 12 13 Data on amputation site and type were reported in 11 trials1–4 6–11 13 and 10 trials,1 2 4 6–11 13 respectively. In these studies, a variety of treatment modalities were used, including: neural block techniques (CPNB and cryoneurolysis) in two trials; neuromodulation therapies (rTMS, ctDCS, and PNS) in three trials; oral medications (amitriptyline, gabapentin, memantine, mexiletine, and morphine) in six trials; and alternative methods (EMS) in one trial. The NMA results, including the MD with 95% CIs and rank probabilities, are illustrated in figure 2. A qualitative summary and network meta-analyses, presented in a league table format, can be found in table 2A,B. Detailed results and relative ranking are listed in online supplemental appendix 7.

    Figure 1
    Figure 1

    PRISMA flow diagram of studies identified and included in this network meta-analysis. CPNB, continuous perineural block; ctDCS, cerebellar transcranial direct current stimulation; EMS, electromagnetic shielding; NM (rTMS), neuromodulation with repetitive transcranial magnetic stimulation; PNS, peripheral nerve stimulation; PO (Amitriptyline), oral administration of amitriptyline; PO (Gabapentin), oral administration of gabapentin; PO (Memantine), oral administration of memantine; PO (Mexiletine), oral administration of mexiletine; PO (Morphine), oral administration of morphine; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; SUCRA, surface under the cumulative ranking curve area.

    View this table:
    Table 1

    Demographic data for the included trials

    Figure 2
    Figure 2

    Network geometry of different interventions for comparisons of changes in pain intensity (A) and adverse event rate (B). SUCRA value as numeric presentation of the overall ranking for all interventions (C–D). The rank would be better with larger value. Forest plots of network estimates were displayed (E–F). Number marked with asterisk indicate significance compared with sham/placebo group. L, low confidence rating; M, moderate confidence rating; SUCRA, surface under the cumulative ranking curve area.

    View this table:
    Table 2

    (A) League table of the changes in pain intensity between different interventions and (B) league table presenting the adverse event rate across different interventions

    Changes in pain intensity

    Twelve trials,1–4 6–13 encompassing 783 participants, were included for analysis of changes in pain intensity. Compared with the sham/placebo group, the summary MD of changes in pain intensity were as follows: −2.90 points (95% CI: −4.62 to –1.18) for rTMS; −1.00 points (95% CI: −3.13 to 1.13) for ctDCS; −1.80 points (95% CI: −3.71 to 0.11) for PNS; −1.50 for CPNB (95% CI: −3.10 to –0.10); 0.23 for cryoneurolysis (95% CI: −1.35 to 1.81); −0.50 for oral amitriptyline (95% CI: −2.68 to 1.68); −1.03 for oral gabapentin (95% CI: −2.29 to 0.23); −0.37 for oral memantine (95% CI: −2.11, 1.37); −0.10 for the oral mexiletine method (95% CI: −1.78 to 1.58); −1.40 for oral morphine (95% CI: −3.05 to –0.25); and 0.20 for the alternative EMS (95% CI: −1.55 to 1.95). A negative MD indicates better pain improvement. The rTMS (SUCRA=94.1%) ranked best for changes in pain intensity, followed by PNS (SUCRA=74.9%) and the CPNB group (SUCRA=70.1%).

    Adverse event rate

    Eight trials,1 3 6 8 10–13 with a total of 466 participants, were included for the analysis of adverse event rate. In comparison with the sham/placebo group, the summary ORs for adverse event rate were: 0.34 (95% CI: 0.01 to 8.44) for cryoneurolysis; 0.60 (95% CI: 0.01 to 32.56) for rTMS; 1.00 (95% CI: 0.02 to 53.89) for ctDCS; 1.17 (95% CI: 0.02 to 63.97) for PNS; 0.68 (95% CI: 0.19 to 2.36) for oral memantine; 1.03 (95% CI: 0.33 to 3.24) for oral mexiletine; 6.04 (95% CI: 2.26 to 16.12) for oral morphine; and 0.90 (95% CI: 0.02 to 47.00) for EMS. An OR less than 1 indicates fewer adverse events. The cryoneurolysis (SUCRA=72.0%) ranked best for adverse event rate, followed by oral memantine (SUCRA=61.4%) and rTMS (SUCRA=59.0%). Reported adverse events for various modalities are detailed in online supplemental appendix 7.2.

    Quality of evidence

    The evidence and summary profile, including GRADE results, is presented in table 3 of online supplemental appendix 11. Most comparisons demonstrated a low to moderate level of confidence regarding changes in pain intensity and the rate of adverse events. Nonetheless, certain comparisons were assigned a very low rating, especially in cases of intransitivity and a high risk of bias.

    View this table:
    Table 3

    Evidence profiles for chronic phantom limb pain treatment in the network meta-analysis

    Inconsistency

    No global inconsistencies (design-by-treatment interaction model) or local inconsistencies (loop approach) were found in changes in pain intensity or adverse event rates (online supplemental appendix 10). The lack of direct comparison data between interventions and the limited closed loops in the network map rendered the results from the side-splitting approach unestimable.

    Publication bias

    In general, the funnel plots displayed a notable degree of symmetry, and Egger’s regression plots did not reveal any significant signs of asymmetry (online supplemental appendix 9).

    Meta-regression

    The meta-regression, which included variables such as the mean initial pain score (above or below 5.8 points), patient age (older or younger than 55 years), time since amputation (more than or less than 2 years), and the predominant amputation site and type (accounting for more than 50%), did not demonstrate statistically significant moderating effects on outcomes related to changes in pain intensity and adverse events (online supplemental appendix 12).

    Sensitivity analysis

    A sensitivity analysis, excluding three trials2 3 8 using imputed pain data (online supplemental appendix 13), showed that rTMS significantly reduced pain compared with placebo or sham (MD=−2.9, 95% CI=−4.42 to –1.38). This method also had fewer adverse events (OR=0.6, 95% CI=0.6 to 0.6) and was top-ranked for pain intensity reduction (SUCRA=95.7%) and low adverse event rates (SUCRA=77.8%).

    Discussion

    This is the first NMA to compare different treatment modalities in terms of efficacy for chronic PLP. Our findings suggest that neuromodulation using rTMS results in a significantly larger pain improvement for chronic PLP than neuromodulation using PNS or nerve blocks with CPNB. Pharmacological treatment with morphine was linked to a significant rise in adverse event rates. The qualitative findings of the NMA are concisely summarized in table 4. The meta-regression analysis, which took into account the baseline pain score, patient age, time since amputation, and amputation site and type, did not influence the results for any of the outcomes. The confidence rating for comparisons varied from very low to moderate, particularly when considering the NMA evidence for changes in pain intensity and adverse event rate.

    View this table:
    Table 4

    Summary findings based on relative rankings from this network meta-analysis

    Chronic PLP stems from complex interactions within the peripheral, spinal, and brain systems.32 A notable cause is the sensorimotor cortex’s misalignment postamputation, leading to heightened neuronal activity.4 8 The extent of cortical reorganization correlates directly with phantom pain severity.3 Additionally, central nervous system adaptations, especially brain reorganization, play a pivotal role in perpetuating the pain.33 Chronic pain, in turn, induces observable brain changes, including gray matter reduction, associated with emotional and cognitive disturbances34 Peripheral elements, such as neuroma development and irregular nerve activity, compound the issue.35 As PLP progresses to chronic neuropathic pain, its intricacies deepen, severely diminishing the patient’s quality of life and rendering treatments like N-methyl D-aspartate (NMDA) antagonists less effective.3 13 There’s a marked disparity between clinical perceptions of PLP prevalence and reality, with current conventional treatments often falling short.6 Comprehensive therapeutic strategies, from pharmaceuticals to innovative techniques, are vital. Notably, methods such as percutaneous PNS, rTMS, and CPNB have shown promise in providing extended relief.2 11 13 Addressing PLP effectively requires a personalized and multifaceted approach, informed by a deep understanding of its roots.36

    In recent literature, neuromodulation modalities have been put forth as potential therapeutic approaches for chronic pain due to their ability to alter maladaptive neuroplasticity and enhance descending inhibitory pathways.16 18 37 A recent NMA suggests that both mirror therapy with phantom exercise and various neuromodulation techniques may be particularly effective in alleviating general PLP. Our NMA further indicated that with the exception of the ctDCS method targeting the cerebellum via cutaneously placed electrodes on the scalp,12 all other neuromodulation interventions presented promising outcomes for chronic PLP alleviation, with none reporting significant adverse events. Particularly noteworthy was rTMS, which uses brief, high-intensity magnetic fields to excite neurons.11 It ranked as the top modality in our NMA and showed an improvement of 2.9 points (95% CI: 1.18 to 4.62) which surpassed the minimal clinically important difference (MCID) threshold set at 1.7 points for chronic PLP36 and 2.0 points for other chronic pains.38 This superior efficacy of rTMS aligns with the theory posited in literature that it potentially restores the motor cortex’s defective areas, possibly through mechanisms involving an increase in serum beta-endorphin levels.11 PNS, which employs flexible open-coil leads placed away from the target nerve using ultrasound guidance,13 39 ranked second. PNS is believed to activate large-diameter fibers effectively, thereby reversing aberrant plasticity and achieving a substantial supraspinal effect.13 Overall, these findings reiterate the conclusions from previous pairwise meta-analyses and clinical studies emphasizing the superiority of neuromodulation modalities in managing chronic pain.16 18

    The administration of peripheral nerve blockade is predominantly used for perioperative management of PLP, frequently targeting the brachial plexus, femoral nerve, and sciatic nerve.3 Traditional nerve blocks, however, often fall short of delivering sustained pain relief for chronic PLP sufferers.2 3 In light of this, continuous perineural infusion and nerve block via cryoneurolysis have been trialed, although with varying outcomes.2 3 Our NMA revealed that nerve block augmented by continuous perineural infusion was notably superior to the control, ranking the third place concerning reductions in pain intensity among all treatments (SUCRA=74.9%). However, the pooled MD in our NMA for pain alleviation by continuous perineural infusion was 1.8 points, falling just above the MCID threshold of 1.7 points set for chronic PLP and under 2.0 for other chronic pains.36 38 Previous study using continuous perineural ropivacaine infusion for 6 days reported that PLP ameliorated shortly post a single ropivacaine injection, maintaining this effect for up to 4 weeks.2 Contrastingly, nerve block using cryoneurolysis, which involves the reversible ablation of peripheral nerves by chilling them with nitrous oxide to approximately −70°C,3 did not exhibit significant pain improvement in our analysis. Earlier studies had similarly reported lackluster outcomes, theorizing that earlier positive results might be attributed to placebo effects, selection biases, or the natural pain resolution process.3 The cryoneurolysis procedure had the lowest ranking for adverse events; however, a previous study emphasized a severe adverse event in a participant who suffered significant weakness in the quadriceps femoris following a transtibial amputation.3 It is worth noting that despite the mixed results for cryoneurolysis, some uncontrolled case series have shown its analgesic benefit for PLP patients.40–42

    Pharmacological interventions have been used historically to treat phantom pain following amputation. These interventions encompass a range of drugs: beta-blockers, calcitonins, anticonvulsants, antidepressants, selective serotonin-reuptake inhibitors, anesthetics, opioids, tramadol, analgesics, NMDA receptor antagonists, non-steroidal anti-inflammatory drugs, and muscle relaxants.15 Despite this variety, for patients suffering from chronic PLP, identifying the optimal pharmacological approach has proven elusive. Most studies have concentrated on opioid analgesics, tricyclic antidepressants, anticonvulsants, NMDAR antagonists, and sodium channel blockers.1 4 7–10 However, our NMA found that none of the following pharmacological treatments: amitriptyline (a tricyclic antidepressant), gabapentin (an anticonvulsant), memantine (an NMDAR antagonist), mexiletine (a sodium channel blocker), or morphine (an opioid analgesic) outperformed the control in terms of pain reduction. Past studies also corroborated these findings, revealing limited efficacy of certain drugs like amitriptyline, memantine, and mexiletine in reducing chronic PLP.1 5 9 10 Furthermore, while some reports suggest morphine’s effectiveness in alleviating chronic PLPs, our NMA contradicts these findings. Our NMA also revealed that morphine, despite its potential benefits for chronic PLP,10 43 carries notable side effects such as nausea, vomiting, dizziness, and drowsiness.10 15 Moreover, the rate of adverse events with morphine was significantly higher compared with placebo (OR=6.04; (95% CI 2.26 to 16.12)) and other pharmacological interventions such as memantine (OR=8.93; (95% CI 1.82 to 43.79)) and mexiletine (OR=5.87; (95% CI 2.19 to 15.70)) (table 2; online supplemental appendix 7.2).

    The EMS system, designed to shield against electromagnetic fields, was believed to work by protecting sensitive nerve endings from environmental electromagnetic disturbances, such as those during thunderstorms.44 45 So far, two RCTs have produced mixed results; one found EMS to be effective,44 while the other found it no better than a placebo.6 In our NMA, EMS performed poorly, ranking below even sham/placebo treatments. This suggests that countering the effects of electromagnetic fields may not play a crucial role in alleviating chronic PLP.

    Limitations

    Our research faces several constraints, most notably the lack of long-term outcome data from the studies reviewed. Of these, only eight trials2–4 6 7 9–11 assessed the effects of interventions beyond 1 month, and just one study2 explored outcomes beyond 6 months. Further RCTs are needed to determine if the immediate benefits persist over time. Additionally, certain interventions, such as neuromodulations (rTMS, ctDCS, and PNS), nerve blocks (CPNB and cryoneurolysis), pharmacological treatments (amitriptyline, mexiletine, and morphine), and the EMS, have each been assessed in just one trial. An analytical approach is thus required for their findings. Confidence in the study outcomes was generally moderate to low, particularly for those with ambiguous evidence. Concerning adverse events, confidence levels were even lower, signaling the need for extra caution. Notably, there was a significant difference in baseline age and pain intensity between the neuromodulation group and others. To prevent overstating the effectiveness of neuromodulations in pain improvement, we downgraded the evidence quality in all related outcomes and acknowledged this inconsistency in our GRADE assessment. Moreover, including cross-over data from the end of the trials tends to underestimate the variance of the treatment effects within these trials, especially when combined with non-cross-over, parallel-group trials. A significant issue highlighted is the absence of standardized methodologies for treating chronic PLP, which might yield inconsistent results. Yet, no inconsistencies between global or local strategies were identified. Finally, the power of our outcome conclusions might be limited due to the inclusion of a comparatively small number of studies.

    Conclusion

    The NMA suggests that neuromodulation using rTMS may be associated with significantly larger pain improvement for chronic PLP. However, the paucity of studies, varying patient characteristics across each trial, and absence of long-term results underscore the necessity for more comprehensive, large-scale RCTs.

    Supplemental material

    Ethics statements

    Patient consent for publication

    References

    1. ↵
      1. Maier C,
      2. Dertwinkel R,
      3. Mansourian N, et al
      . Efficacy of the NMDA-receptor antagonist memantine in patients with chronic phantom limb pain--results of a randomized double-blinded, placebo-controlled trial. Pain 2003;103:277–83. doi:10.1016/S0304-3959(02)00456-6
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Ilfeld BM,
      2. Khatibi B,
      3. Maheshwari K, et al
      . Ambulatory continuous peripheral nerve blocks to treat postamputation phantom limb pain: A multicenter, randomized, quadruple-masked, placebo-controlled clinical trial. Pain 2021;162:938–55. doi:10.1097/j.pain.0000000000002087
      OpenUrlPubMed
    3. ↵
      1. Ilfeld BM,
      2. Smith CR,
      3. Turan A, et al
      . Ultrasound-guided Percutaneous Cryoneurolysis to Treat Chronic Postamputation Phantom Limb Pain: A Multicenter Randomized Controlled Trial. Anesthesiology 2023;138:82–97. doi:10.1097/ALN.0000000000004429
      OpenUrl
    4. ↵
      1. Smith DG,
      2. Ehde DM,
      3. Hanley MA, et al
      . Efficacy of gabapentin in treating chronic phantom limb and residual limb pain. J Rehabil Res Dev 2005;42:645–54. doi:10.1682/jrrd.2005.05.0082
      OpenUrlCrossRefPubMed
    5. ↵
      1. Wiech K,
      2. Kiefer R-T,
      3. Töpfner S, et al
      . A placebo-controlled randomized crossover trial of the N-methyl-D-aspartic acid receptor antagonist, memantine, in patients with chronic phantom limb pain. Anesth Analg 2004;98:408–13. doi:10.1213/01.ANE.0000096002.53818.BD
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Hsiao A-F,
      2. York R,
      3. Hsiao I, et al
      . A randomized controlled study to evaluate the efficacy of noninvasive limb cover for chronic phantom limb pain among veteran amputees. Arch Phys Med Rehabil 2012;93:617–22. doi:10.1016/j.apmr.2011.11.021
      OpenUrlPubMed
    7. ↵
      1. Bone M,
      2. Critchley P,
      3. Buggy DJ
      . Gabapentin in postamputation phantom limb pain: A randomized, double-blind, placebo-controlled, cross-over study. Reg Anesth Pain Med 2002;27:481–6. doi:10.1053/rapm.2002.35169
      OpenUrlAbstract/FREE Full Text
    8. ↵
      1. Schwenkreis P,
      2. Maier C,
      3. Pleger B, et al
      . NMDA-mediated mechanisms in cortical excitability changes after limb amputation. Acta Neurol Scand 2003;108:179–84. doi:10.1034/j.1600-0404.2003.00114.x
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Robinson LR,
      2. Czerniecki JM,
      3. Ehde DM, et al
      . Trial of amitriptyline for relief of pain in amputees: results of a randomized controlled study. Arch Phys Med Rehabil 2004;85:1–6. doi:10.1016/s0003-9993(03)00476-3
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Wu CL,
      2. Agarwal S,
      3. Tella PK, et al
      . Morphine versus mexiletine for treatment of postamputation pain: A randomized, placebo-controlled, crossover trial. Anesthesiology 2008;109:289–96. doi:10.1097/ALN.0b013e31817f4523
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Ahmed MA,
      2. Mohamed SA,
      3. Sayed D
      . Long-term antalgic effects of repetitive transcranial magnetic stimulation of motor cortex and serum beta-endorphin in patients with phantom pain. Neurol Res 2011;33:953–8. doi:10.1179/1743132811Y.0000000045
      OpenUrlCrossRefPubMed
    12. ↵
      1. Bocci T,
      2. De Carolis G,
      3. Ferrucci R, et al
      . Cerebellar Transcranial Direct Current Stimulation (ctDCS) Ameliorates Phantom Limb Pain and Non-painful Phantom Limb Sensations. Cerebellum 2019;18:527–35. doi:10.1007/s12311-019-01020-w
      OpenUrl
    13. ↵
      1. Gilmore CA,
      2. Ilfeld BM,
      3. Rosenow JM, et al
      . Percutaneous 60-day peripheral nerve stimulation implant provides sustained relief of chronic pain following amputation: 12-month follow-up of a randomized, double-blind, placebo-controlled trial. Reg Anesth Pain Med 2019:rapm-2019-100937. doi:10.1136/rapm-2019-100937
    14. ↵
      1. Flor H
      . Phantom-limb pain: characteristics, causes, and treatment. Lancet Neurol 2002;1:182–9. doi:10.1016/s1474-4422(02)00074-1
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Alviar MJM,
      2. Hale T,
      3. Dungca M
      . Pharmacologic interventions for treating phantom limb pain. Cochrane Database Syst Rev 2016;10:CD006380. doi:10.1002/14651858.CD006380.pub3
    16. ↵
      1. Pacheco-Barrios K,
      2. Meng X,
      3. Fregni F
      . Neuromodulation Techniques in Phantom Limb Pain: A Systematic Review and Meta-analysis. Pain Med 2020;21:2310–22. doi:10.1093/pm/pnaa039
      OpenUrl
    17. ↵
      1. Laloo R,
      2. Ambler GK,
      3. Locker D, et al
      . Systematic Review and Meta-Analysis of the Effect of Perineural Catheters in Major Lower Limb Amputations. Eur J Vasc Endovasc Surg 2021;62:295–303. doi:10.1016/j.ejvs.2021.03.008
      OpenUrlPubMed
    18. ↵
      1. Navarro-López V,
      2. Del-Valle-Gratacós M,
      3. Fernández-Vázquez D, et al
      . Transcranial direct current stimulation in the management of phantom limb pain: A systematic review of randomized controlled trials. Eur J Phys Rehabil Med 2022;58:738–48. doi:10.23736/S1973-9087.22.07439-1
      OpenUrl
    19. ↵
      1. Guémann M,
      2. Olié E,
      3. Raquin L, et al
      . Effect of mirror therapy in the treatment of phantom limb pain in amputees: A systematic review of randomized placebo-controlled trials does not find any evidence of efficacy. Eur J Pain 2023;27:3–13. doi:10.1002/ejp.2035
      OpenUrl
    20. ↵
      1. Neumüller J,
      2. Lang-Illievich K,
      3. Brenna CTA, et al
      . Calcitonin in the Treatment of Phantom Limb Pain: A Systematic Review. CNS Drugs 2023;37:513–21. doi:10.1007/s40263-023-01010-x
      OpenUrl
    21. ↵
      1. Wang J,
      2. Fan J,
      3. Gc R, et al
      . Comparative Effects of Interventions on Phantom Limb Pain: A Network Meta-Analysis. World Neurosurgery 2023;170:e45–56. doi:10.1016/j.wneu.2022.10.060
      OpenUrl
    22. ↵
      1. Limakatso K,
      2. Parker R
      . Treatment Recommendations for Phantom Limb Pain in People with Amputations: An Expert Consensus Delphi Study. PM R 2021;13:1216–26. doi:10.1002/pmrj.12556
      OpenUrlPubMed
    23. ↵
      1. Follmann D,
      2. Elliott P,
      3. Suh I, et al
      . Variance imputation for overviews of clinical trials with continuous response. J Clin Epidemiol 1992;45:769–73. doi:10.1016/0895-4356(92)90054-q
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Sterne JAC,
      2. Savović J,
      3. Page MJ, et al
      . RoB 2: A revised tool for assessing risk of bias in randomised trials. BMJ 2019;366:l4898. doi:10.1136/bmj.l4898
    25. ↵
      1. Brignardello-Petersen R,
      2. Bonner A,
      3. Alexander PE, et al
      . Advances in the GRADE approach to rate the certainty in estimates from a network meta-analysis. J Clin Epidemiol 2018;93:36–44. doi:10.1016/j.jclinepi.2017.10.005
      OpenUrlCrossRefPubMed
    26. ↵
      1. Puhan MA,
      2. Schünemann HJ,
      3. Murad MH, et al
      . A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. BMJ 2014;349:g5630. doi:10.1136/bmj.g5630
    27. ↵
      1. Mbuagbaw L,
      2. Rochwerg B,
      3. Jaeschke R, et al
      . Approaches to interpreting and choosing the best treatments in network meta-analyses. Syst Rev 2017;6:79. doi:10.1186/s13643-017-0473-z
    28. ↵
      1. Dias S,
      2. Welton NJ,
      3. Caldwell DM, et al
      . Checking consistency in mixed treatment comparison meta-analysis. Stat Med 2010;29:932–44. doi:10.1002/sim.3767
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Higgins JPT,
      2. Jackson D,
      3. Barrett JK, et al
      . Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies. Res Synth Methods 2012;3:98–110. doi:10.1002/jrsm.1044
      OpenUrlCrossRefPubMed
    30. ↵
      1. Boonstra AM,
      2. Schiphorst Preuper HR,
      3. Balk GA, et al
      . Cut-off points for mild, moderate, and severe pain on the visual analogue scale for pain in patients with chronic musculoskeletal pain. Pain 2014;155:2545–50. doi:10.1016/j.pain.2014.09.014
      OpenUrlCrossRefPubMed
    31. ↵
      1. Ephraim PL,
      2. Wegener ST,
      3. MacKenzie EJ, et al
      . Phantom pain, residual limb pain, and back pain in amputees: results of a national survey. Arch Phys Med Rehabil 2005;86:1910–9. doi:10.1016/j.apmr.2005.03.031
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. Ehde DM,
      2. Czerniecki JM,
      3. Smith DG, et al
      . Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation. Arch Phys Med Rehabil 2000;81:1039–44. doi:10.1053/apmr.2000.7583
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Moseley GL,
      2. Flor H
      . Targeting cortical representations in the treatment of chronic pain: A review. Neurorehabil Neural Repair 2012;26:646–52. doi:10.1177/1545968311433209
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Seminowicz DA,
      2. Wideman TH,
      3. Naso L, et al
      . Effective Treatment of Chronic Low Back Pain in Humans Reverses Abnormal Brain Anatomy and Function. J Neurosci 2011;31:7540–50. doi:10.1523/JNEUROSCI.5280-10.2011
      OpenUrlAbstract/FREE Full Text
    35. ↵
      1. Cohen SP,
      2. Gilmore CA,
      3. Rauck RL, et al
      . Percutaneous Peripheral Nerve Stimulation for the Treatment of Chronic Pain Following Amputation. Mil Med 2019;184:e267–74. doi:10.1093/milmed/usz114
      OpenUrl
    36. ↵
      1. Kikkert S,
      2. Johansen-Berg H,
      3. Tracey I, et al
      . Reaffirming the link between chronic phantom limb pain and maintained missing hand representation. Cortex 2018;106:174–84. doi:10.1016/j.cortex.2018.05.013
      OpenUrlCrossRefPubMed
    37. ↵
      1. Akyuz G,
      2. Giray E
      . Noninvasive neuromodulation techniques for the management of phantom limb pain: A systematic review of randomized controlled trials. Int J Rehabil Res 2019;42:1–10. doi:10.1097/MRR.0000000000000317
      OpenUrl
    38. ↵
      1. Farrar JT,
      2. Young JP,
      3. LaMoreaux L, et al
      . Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain 2001;94:149–58. doi:10.1016/S0304-3959(01)00349-9
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Rauck RL,
      2. Cohen SP,
      3. Gilmore CA, et al
      . Treatment of post-amputation pain with peripheral nerve stimulation. Neuromodulation 2014;17:188–97. doi:10.1111/ner.12102
      OpenUrlCrossRefPubMed
    40. ↵
      1. Moesker AA,
      2. Karl HW,
      3. Trescot AM
      . Treatment of phantom limb pain by cryoneurolysis of the amputated nerve. Pain Pract 2014;14:52–6. doi:10.1111/papr.12020
      OpenUrl
    41. ↵
      1. Neumann V,
      2. O’Connor RJ,
      3. Bush D
      . Cryoprobe treatment: an alternative to phenol injections for painful neuromas after amputation. AJR Am J Roentgenol 2008;191:W313; author reply W314. doi:10.2214/AJR.08.1371
    42. ↵
      1. Prologo JD,
      2. Gilliland CA,
      3. Miller M, et al
      . Percutaneous Image-Guided Cryoablation for the Treatment of Phantom Limb Pain in Amputees: A Pilot Study. J Vasc Interv Radiol 2017;28:24–34. doi:10.1016/j.jvir.2016.09.020
      OpenUrl
    43. ↵
      1. Huse E,
      2. Larbig W,
      3. Flor H, et al
      . The effect of opioids on phantom limb pain and cortical reorganization. Pain 2001;90:47–55. doi:10.1016/s0304-3959(00)00385-7
      OpenUrlCrossRefPubMedWeb of Science
    44. ↵
      1. Fisher K,
      2. Oliver S,
      3. Sedki I, et al
      . The effect of electromagnetic shielding on phantom limb pain: A placebo-controlled double-blind crossover trial. Prosthet Orthot Int 2016;40:350–6. doi:10.1177/0309364614568409
      OpenUrlCrossRefPubMed
    45. ↵
      1. Giummarra MJ,
      2. Georgiou-Karistianis N,
      3. Nicholls MER, et al
      . The menacing phantom: what pulls the trigger? Eur J Pain 2011;15:691. doi:10.1016/j.ejpain.2011.01.005
      OpenUrlPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Contributors S-MC, J-CW and C-AS: conceived and designed the study. SMC, J-CW, C-AS, S-CL and C-RL: acquired the data. SMC, J-CW, S-CL, C-RL, P-TW, F-CK, C-JF, Y-KT, K-LH and P-CL, CAS: analyzed and interpreted the data. SMC, J-CW and C-AS: drafted the article. All authors read and approved the final manuscript.

    • Funding This research received support from grants NSTC 111-2314-B-006-117 and NCKUH-11303051. Additionally, the research was supported in part by Higher Education Sprout Project, Ministry of Education to the Headquarters of University Advancement at National Cheng Kung University (NCKU).

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.