1. Introduction
Amputation is a profound physical and psychological trauma with a high global incidence. Beyond the loss of mobility, patients frequently face persistent pain, increased dependency, diminished quality of life, and a higher prevalence of depression and other mental health disorders [1].
Pain remains the most prevalent clinical symptom in the postoperative period following amputation [2]. Most patients experience two distinct manifestations: residual limb pain (RLP), which typically subsides after tissue healing, and phantom limb pain (PLP), which persists beyond the healing phase and can become a lifelong condition [3].
Effective management requires the healthcare team to periodically assess and differentiate these clinical forms. Prompt and individualized treatment is essential to prevent the transition from acute to chronic pain, thereby promoting patient well‐being and functional recovery [1, 4]. Current strategies to alleviate post‐amputation pain include specialized surgical techniques, regional analgesia, pharmacological interventions, psychotherapy, and supportive care such as stump dressings, cryotherapy, therapeutic massage, mirror therapy, and patient education regarding prosthesis care [5].
Despite this multidisciplinary repertoire, a consensus on the most effective pain control protocol remains elusive in the scientific literature, partly due to a pathophysiology that is not yet fully understood [5–10]. Commonly utilized pharmacological therapies include opioid and nonopioid analgesics, nonsteroidal anti‐inflammatory drugs (NSAIDs), epidural or perineural anesthetics, N‐methyl‐D‐aspartate (NMDA) receptor antagonists, gabapentinoids, antidepressants, and calcitonin [2, 5, 9].
However, the evidence regarding the efficacy of these treatments is often divergent, and the wide range of associated adverse effects poses significant risks to patient safety [1, 2, 5, 9, 11]. Furthermore, there is a lack of high‐level evidence synthesis that meticulously compares these therapies to establish their comparative short‐ and long‐term benefits.
Given that postamputation pain causes substantial suffering and that medications are often the primary intervention in clinical settings, bridging these literature gaps is critical. Since the appropriate selection of therapy directly impacts symptom control and clinical outcomes, this study seeks to provide a robust synthesis of the existing evidence [10–12].
The aim of this study is to evaluate the efficacy and safety of pharmacological therapies for pain management in patients undergoing amputation. By conducting a comprehensive systematic review, this research seeks to support evidence‐based therapeutic choices, stimulate academic debate, and generate knowledge that clarifies the clinical impact of these interventions. Furthermore, the inclusion of a network meta‐analysis (NMA) allows for the comparative ranking of treatments, providing a hierarchical evidence base to support clinical decision‐making and optimize patient care.
2. Materials and Methods
This study consists of a systematic review and NMA conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions [13]. The reporting follows the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses for NMA (PRISMA‐NMA) statement [14] (Supporting File 1). The study protocol was prospectively registered in the PROSPERO database under the identifier CRD42024500600 (Supporting File 2).
2.1. Search Strategy
To develop the search strategy, relevant descriptors were identified and selected based on the PICOS framework, defined as follows:•Population (P): adults and older adults (aged ≥ 18 years) with limb amputation.•Intervention (I): pharmacological therapies.•Control (C): placebo or alternative therapies.•Outcomes (O): pain management.•Study design (S): systematic reviews with or without meta‐analysis.
Search terms were selected according to the Medical Subject Headings (MeSH) and Health Sciences Descriptors (DeCS). Boolean operators (AND, OR) were applied to combine descriptors into specific strings optimized for each database. Preliminary searches were conducted across all databases to refine these combinations according to the specific indexing features of each source, as per Table 1.
Following the finalization of the search strings, the systematic search was performed by two independent, calibrated researchers to ensure methodological rigor.
2.2. Eligibility Criteria
We included systematic reviews, with or without meta‐analysis, evaluating pharmacological interventions for pain management in patients aged 18 years or older (adults and older adults) who had undergone amputation. No restrictions were placed on publication date or language, provided the full text was available.
Studies were excluded if they (1) were duplicates; (2) did not directly address the research question; (3) and were unavailable in full‐text format or had restricted access. Furthermore, gray literature including editorials, manuals, dissertations, theses, study protocols, pilot studies, conference abstracts, and advertisements was excluded to ensure the inclusion of peer‐reviewed scientific research only.
2.3. Search Strategy and Study Selection
The search was conducted in September 2024 across the following databases: Cochrane Library, LILACS, PubMed/MEDLINE, SciELO, ScienceDirect, Scopus, CINAHL, Web of Science, and Google Scholar. Most data sources were accessed through the Coordination for the Improvement of Higher Education Personnel (CAPES) Periodicals Portal via the Federated Academic Community (CAFe), facilitated by the Federal University of Rio Grande do Norte (UFRN). The Cochrane Library and Google Scholar were accessed directly through their respective official platforms.
To ensure methodological rigor and minimize the risk of bias, errors, or discrepancies, all stages of the review including the search, study selection, and data extraction were performed by two independent, calibrated researchers.
The Rayyan systematic review software was utilized to identify and remove duplicate records. Subsequently, the screening of titles and abstracts, followed by full‐text evaluation, was conducted independently by the two reviewers. Any persistent discrepancies were resolved through consensus or by a third senior researcher.
2.4. Quality Assessment
The methodological quality and reporting integrity of the included systematic reviews were evaluated using the Assessment of Multiple Systematic Reviews 2 (AMSTAR 2) tool. This instrument comprises 16 items that allow for a comprehensive appraisal of systematic reviews of both randomized and nonrandomized studies. Based on the assessment, the reviews were categorized into four levels of confidence: high, moderate, low, or critically low [15].
2.5. Data Abstraction
Data from the selected studies were extracted independently by two researchers and managed using Microsoft Excel and Google Docs. The extracted information included bibliographic and sociodemographic data (e.g., author, year, and country); clinical variables related to the amputation (e.g., level of amputation and etiology); pharmacological intervention details (e.g., drug class, dosage, administration route); and pain‐related outcomes (e.g., pain intensity scales for residual and PLP).
2.6. Synthesis
A narrative synthesis was first conducted to summarize the characteristics, interventions, and clinical outcomes of all included studies. For the quantitative synthesis, a NMA was performed using a frequentist approach via the “netmeta” package in the R software environment (Version 4.0.3; R Core Team, 2020). Efficacy and safety estimates were calculated using frequencies, means, standard deviations, and 95% confidence intervals (CIs) derived from the primary randomized controlled trials (RCTs).
Prior to the analysis, the network geometry was inspected to determine connectivity. In cases where the network was disconnected, subnetworks were identified and analyzed separately. Subnetworks consisting of only a single study were excluded from the meta‐analysis due to the impossibility of conducting comparative estimates.
3. Results
3.1. Study Selection
The initial database search yielded 2352 records. After removing duplicates, 841 records remained for title and abstract screening. Following this preliminary phase, 43 articles were selected for full‐text assessment.
Upon comprehensive full‐text review and application of the eligibility criteria, 17 systematic reviews were deemed eligible for inclusion in this study. From this final sample, seven reviews comprising RCTs provided sufficient data for the NMA. The complete process of identification, screening, and selection is detailed in the PRISMA flow diagram (Figure 1).
3.2. Synthesis
A summary of the primary characteristics of the 17 included studies is presented in Table 2. Across these studies, a total of 17,011 participants were evaluated, with ages ranging from 19 to 92 years.
Notably, none of the included studies reported specific data regarding the ethnicity or racial background of the participants. This represents a significant gap in the primary literature concerning the demographic representation of the study populations.
The 17 systematic reviews were published between 2016 and 2024, originating from Canada (3), the United Kingdom (3), Australia (2), Spain (2), China (2), South Africa (1), Brazil (1), the United States (1), Finland (1), and New Zealand (1). Six studies included a meta‐analysis, and seven were composed exclusively of RCTs. The population comprised patients with both upper and lower limb amputations.
Seven primary etiologies for amputation were reported: vascular disorders, neoplasms, trauma, infectious diseases, diabetes mellitus, chronic pain, and penetrating injuries (stab wounds). A total of 29 different instruments were identified for pain assessment. The most frequent were: visual analog scale (VAS): 11 studies (64.7%); numeric rating scale (NRS): 10 studies (58.8%); and McGill Pain Questionnaire (MPQ): 5 studies (29.4%).
Notably, four studies (23.5%) did not clearly specify the pain assessment method utilized.
Regarding therapeutic strategies, 58 distinct interventions were identified, categorized as follows: Pharmacological therapies: 25 (43.1%); complementary and integrative practices: 21 (36.2%); standard or routine care: 8 (13.8%); surgical therapies: 6 (10.3%), including perioperative anesthetics and targeted muscle reinnervation (TMR); physiotherapy: 5 (8.6%); and multidisciplinary treatment: 1 (1.8%).
Standard or routine care encompassed a multiprofessional approach involving pharmacological and nonpharmacological interventions, perioperative care, complication prevention, and psychosocial support [16–18].
The primary clinical objective was pain treatment (n = 14; 82.3%), followed by prevention (n = 2; 11.8%) and a combination of both (n = 1; 5.9%). Interventions were most frequently applied in the postoperative period (n = 13; 76.5%), while others spanned the intraoperative/postoperative (n = 2; 11.8%), strictly intraoperative (n = 1; 5.9%), or all surgical phases (n = 1; 5.9%).
Regarding pain characteristics, PLP was the most prevalent, with 10 studies (58.8%) focusing exclusively on this condition. Specific dosages and administration routes for pharmacological therapies were not detailed due to significant data heterogeneity and missing information in the primary reports, despite unsuccessful attempts to retrieve these data by contacting the authors.
3.3. Methodological Quality Assessment
The AMSTAR 2 assessment present in Table 3 revealed that among the 17 included reviews, 7 (41.2%) were rated as “moderate,” 7 (41.2%) were rated as “critically low,” and 3 (17.6%) were rated as “low” regarding overall confidence in the results.
The included reviews performed poorly across three critical domains of the AMSTAR 2 tool: (1) Protocol Registration (Item 2): Failure to register a protocol before the commencement of the review, increasing the risk of selective reporting bias; (2) Justification for Exclusions (Item 7): Absence of a detailed list of excluded studies with the specific rationales for their removal; and (3) Risk of Bias Integration (Item 13): Inadequate consideration of the Risk of Bias (RoB) of primary studies when interpreting and discussing the synthesis of results.
3.4. NMA
Seven systematic reviews of clinical trials were identified, comprising 72 primary studies. Of these, 24 were duplicates and were excluded, leaving 48 studies for full‐text review and screening. Ultimately, 17 primary studies were included in the intervention networks.
Two outcomes were analyzed, forming subgroups: PLP (n = 16 studies) and RLP (n = 3 studies). For pain assessment, numerical scales with continuous data were used, as other types of scales with qualitative assessments precluded the performance of a meta‐analysis.
The meta‐analysis for each outcome was conducted using forest plots and combinations of direct and indirect comparisons. The variables analyzed in the plots included the total number of participants, the mean value of the outcome, and the corresponding standard deviation, all reported for both groups. The 95% CIs were also considered.
For both PLP and RLP outcomes related to amputations, networks were constructed based on the experimental and control therapy groups within the RCTs. For the PLP outcome, five subnetworks were identified. The studies evaluated the following interventions: transcranial magnetic stimulation (TMS) applied to the motor cortex compared with stimulation applied to the vertex; lidocaine versus botulinum toxin A; and phantom exercises compared with a general exercise program.
The PLP outcome also allowed for the analysis of two subnetworks. In the first PLP subnetwork, each treatment was compared with placebo, as all studies in this subgroup evaluated therapies (amitriptyline, gabapentin, memantine, morphine, and TMS) against placebo. Figure 2 presents the estimates derived from both direct and indirect comparisons.
No statistically significant differences were observed among the therapies, and a treatment ranking is also provided.
Regarding the ranking of therapies for PLP in the first subnetwork, the analysis indicated that the treatment with the highest probability of achieving the best patient outcome was TMS (p‐score = 0.6949), followed by memantine (p‐score = 0.5309). Gabapentin ranked lowest, with a p‐score of 0.2337.
The adjusted model for the second subnetwork evaluated mirror therapy in comparison with other treatments, as shown in Figure 3. No statistically significant differences were identified in the reduction in PLP among the treatments in either direct or indirect comparisons. According to the treatment ranking for this subnetwork, mirror therapy showed the highest probability of pain reduction (p‐score = 0.6058), whereas covered mirror therapy, used as a placebo in the clinical trials, demonstrated the lowest probability (p‐score = 0.3426).
For the second outcome, RLP, two networks were identified: one comprising a single study and the other comprising two studies. Consequently, only one subnetwork was suitable for the analysis of the RLP outcome.
The analysis of this RLP subnetwork, which included the pharmacological therapies amitriptyline and gabapentin, did not demonstrate statistically significant differences compared with placebo in either direct or indirect comparisons, as illustrated in Figure 4.
Although no statistically significant differences were detected among the therapies for RLP, the treatment ranking placed gabapentin first (p‐score = 0.8082). Finally, well‐connected network diagrams were generated for the PLP and RLP subnetworks, including all interventions, as shown in Figure 5. Solid lines represent the direct comparisons, whereas dashed lines indicate the available indirect comparisons between pairs of interventions.
4. Discussion
This study provides a comprehensive evaluation of pharmacological therapies for pain management in postamputation patients, offering insights into their efficacy, safety profiles, and the clinical limitations that impact patient outcomes.
Our findings reinforce that the most affected demographic consists of young and older adult males [19, 20]. Global Burden of Disease data across 204 countries identify a similar trend: men aged 20–60 years are disproportionately affected by amputation due to both clinical conditions and external causes [21]. This gender disparity is partly attributed to higher male representation in physical labor and greater exposure to severe occupational injuries [21].
Furthermore, over 60% of traumatic amputations result from traffic accidents involving young men. Other contributing factors include increased exposure to violence, lifestyle‐related risks, and chronic conditions such as Type 1 diabetes and infectious diseases [19, 21].
In the elderly population, amputation rates increase alongside physiological decline, which elevates the risk of falls, osteoporosis, and the prevalence of metabolic comorbidities [19].
Postamputation pain is a near‐universal experience. Our synthesis confirms that PLP and RLP are the most frequently investigated conditions. While these manifestations are often treated postoperatively, evidence suggests that preventive measures initiated before or during surgery are critical to minimizing symptom intensity.
Effective pain management must begin in the preoperative phase, considering the underlying etiology, the characteristics of pre‐existing pain, and the specific amputation site [22]. Within this context, the nursing team plays a pivotal role. Nurses are often the first to receive patient complaints and are responsible for implementing systematized care plans, leading multidisciplinary actions, and continuously re‐evaluating the effectiveness of interventions [12, 23, 24].
Regarding pharmacological prevention, previous research has highlighted the protective effect of epidural therapies (e.g., bupivacaine combined with fentanyl or calcitonin) administered 48 h prior to the procedure. Patients receiving such preventive measures often report lower pain intensity even when symptoms do manifest [25].
Our review observed that preventive approaches remain underexplored in the literature. Most pharmacological interventions are reactive, initiated in the postoperative period for both RLP and PLP. The identified pharmacological repertoire is vast, including opioids, NMDA antagonists, gabapentinoids (anticonvulsants), antidepressants, and botulinum toxin A, often used in conjunction with nonpharmacological strategies and routine institutional care.
Despite this variety, our NMA revealed a significant lack of statistical superiority among these agents. This suggests that while many drugs are utilized in clinical practice, their comparative efficacy remains unproven, highlighting a critical need for standardized protocols and higher‐quality primary studies.
It is important to note that not all therapies identified in this overview were included in the NMA. To ensure the highest level of evidence and internal validity, only primary RCTs were selected for the quantitative synthesis of therapeutic efficacy.
Opioids and their derivatives including morphine, meperidine, codeine, fentanyl, and pethidine remained the most frequently cited drug class. While opioids are staples in the management of major surgical pain, they present a well‐documented clinical paradox [26]. Despite providing potent acute analgesia, their extensive side‐effect profile including constipation, fatigue, vertigo, urinary retention, pruritus, and life‐threatening respiratory depression significantly compromises patient safety [7].
Furthermore, prolonged opioid use (defined as continuous prescription between 90 and 180 days postamputation) poses severe long‐term risks, such as substance dependence, hyperalgesia, depression, and social withdrawal [12, 24, 26, 27]. These risks are particularly concerning for patients with PLP, a chronic condition that may persist for years. Consequently, developing “opioid‐sparing” strategies remains a critical challenge for the scientific community.
Our findings highlight bupivacaine and fentanyl as the most utilized anesthetics. The literature consistently supports the safety and efficacy of the epidural or perineural approach. Continuous administration via catheters for 72 h postoperatively has been shown to effectively reduce acute pain intensity and may prevent the transition to chronic pain states [6, 25, 28].
Promising alternatives to traditional pharmacotherapy include TMR [18, 29] and nonpharmacological interventions such as mirror therapy, virtual reality, and motor imagery, which were featured in our PLP subnetworks.
Conversely, several pharmacological classes including botulinum toxin A, antidepressants, anticonvulsants, and barbiturates showed limited and often conflicting evidence regarding their efficacy for postamputation pain. Notably, there is currently insufficient evidence to support the routine use of anticholinergics, antispasmodics, or adrenergic agonists, as these lack high‐quality RCTs to justify their clinical administration [6, 9, 30].
Our NMA identified amitriptyline and gabapentin within the RLP and PLP subnetworks. Amitriptyline, a tricyclic antidepressant, modulates serotonin, and norepinephrine levels to assist in pain control. Despite its widespread use for chronic neuropathic pain—common in traumatic amputations, our findings align with previous reports showing no significant superiority over placebo for PLP [7].
Similarly, gabapentin, an anticonvulsant targeting voltage‐gated calcium channels, presented a dichotomous result in our ranking: It placed last for PLP but first for RLP. This discrepancy underscores the controversial nature of gabapentin in this population. Its integration into clinical practice should be approached with caution due to inconsistent efficacy data and a notable profile of adverse effects [7, 9].
Regarding memantine, although it ranked second in our PLP treatment hierarchy, it failed to achieve statistical significance in either direct or indirect comparisons. This supports earlier systematic reviews where daily administration of 30 mg for up to 4 weeks showed no clinically significant reduction in pain intensity [7].
A consistent theme across the reviewed evidence is the shift toward multimodal pain management. Integrating diverse analgesic mechanisms can synergistically enhance anesthetic effects while reducing opioid consumption, hospital stays, and overall healthcare costs.
A successful example of this approach identified in our study is the combination of simple analgesia (e.g., acetaminophen) with anti‐inflammatory agents (e.g., methylprednisolone, parecoxib, or NSAIDs) during the perioperative period [6, 25]. Furthermore, combining pharmacological agents with nonpharmacological therapies—such as mirror therapy or virtual reality represents a promising, holistic frontier for pain management [10, 16, 17, 31].
However, the lack of homogeneity in current study populations limits definitive conclusions. Future RCTs must prioritize standardized protocols to test the long‐term efficacy of these combined therapeutic strategies in the amputee population [32].
Within the PLP intervention networks, two nonpharmacological therapies showed notable ranking probabilities. In the first subnetwork, TMS emerged as the intervention with the highest probability of pain reduction, while mirror therapy led the second subnetwork. Conversely, the lack of studies representing nonpharmacological interventions for RLP identifies a significant gap in the literature and a priority for future research.
TMS facilitates the investigation of neural plasticity and cortical reorganization, providing a framework to modulate maladaptive neuroplasticity and enhance descending inhibitory pathways. When administered repetitively, TMS can adjust cortical excitability, showing promise as a neuromodulatory tool for pain management [33]. However, its therapeutic efficacy for PLP remains inconsistent; while some reviews report improvements in pain intensity, anxiety, and depressive symptoms, others found no significant difference compared to placebo or noted that benefits became negligible after 30 days [34–36].
The primary barriers to establishing the efficacy and safety of TMS for PLP include limited research output, small sample sizes, and a lack of standardized parameters—such as stimulation intensity, session frequency, and pulse count. As observed in current reviews, these technical parameters are crucial for treatment success but remain highly heterogeneous across studies [33–35].
Regarding mirror therapy, five systematic reviews evaluated its application, including variations such as covered mirrors, virtual reality, augmented reality, and gamified exercise programs [10, 16, 17, 31]. Mirror therapy is widely recognized for its clinical feasibility and ease of self‐administration following proper instruction. By reflecting the healthy limb, the therapy re‐establishes a coherent body image, which may alleviate phantom sensations and pain by resolving sensorimotor incongruence [37, 38].
Despite its potential, reviews highlight significant diversity in the implementation of these practices and the instruments used to quantify their effects. This methodological variance hinders the classification of mirror therapy as a definitive “safe and effective” intervention [37, 39]. To overcome this, the development, validation, and dissemination of standardized protocols are essential to ensure replicability in future high‐quality RCTs for the amputee population.
5. Limitations
It is essential to acknowledge several limitations that temper the findings of this review. First, there is a profound lack of long‐term longitudinal data; only one included review identified a follow‐up period of up to 2 years in its primary RCTs [6]. In twelve of the seventeen reviews, follow‐up durations were not clearly specified.
Consequently, it remains uncertain whether the immediate analgesic benefits and safety profiles of the evaluated therapies persist over time.
Furthermore, the statistical power of certain comparisons was constrained by the fact that several pharmacological agents including botulinum toxin A, dextromethorphan, benztropine mesylate, meperidine, methylprednisolone, and clonidine were evaluated in only a single RCT. This scarcity of primary data precludes robust meta‐analytical synthesis for these specific interventions.
Moreover, the diversity of outcome metrics ranging from various pain scales to imaging findings prevented the inclusion of several potentially relevant studies in the quantitative synthesis, potentially.
Finally, most studies failed to report clinically relevant secondary outcomes, such as quality of life, sleep patterns, or hemodynamic stability (vital signs), which are vital for a holistic understanding of patient recovery. Lastly, as this review focused exclusively on adult and geriatric populations, the findings cannot be generalized to pediatric patients, who may present distinct pathophysiological responses to postamputation pain.
6. Conclusions
This systematic review and NMA identified a broad spectrum of pharmacological and nonpharmacological interventions for managing PLP and RLP. Most strategies are reactive, focused on the postoperative period, with preventive approaches remaining significantly underexplored.
The pharmacological landscape is diverse, encompassing opioids, anesthetics, NMDA antagonists, gabapentinoids, and antidepressants, among others. However, all identified pharmacological therapies were associated with adverse effects, posing varying degrees of risk to patient safety. Furthermore, the NMA revealed no statistically significant differences in efficacy across the evaluated therapies in either direct or indirect comparisons.
While TMS and mirror therapy achieved the highest relative rankings (P‐scores), their clinical superiority remains unconfirmed due to high study heterogeneity.
Given these results and the identified methodological limitations, current scientific evidence is insufficient to confirm the efficacy and safety of specific pharmacological therapies for pain management in postamputation patients. Many interventions yield inconsistent results, and primary studies often suffer from small, heterogeneous samples and a lack of standardized outcome metrics. Critically, essential aspects of patient well‐being—such as quality of life, sleep patterns, and hemodynamic stability—are rarely addressed in the literature.
Therefore, future research must prioritize methodological rigor and standardized protocols to ensure replicability. Prospective studies should adopt a holistic approach, considering surgical etiology, patient history, socioeconomic factors, and the pharmacokinetics of combined therapies. Specifically, the development of robust, multimodal strategies that integrate pharmacological and nonpharmacological interventions is essential to optimize pain control and enhance the recovery of patients undergoing amputation.
Funding
This research did not receive any specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting Information
Additional supporting information can be found online in the Supporting Information section.
Supporting information