Evidence Mapping Based on Systematic Reviews of Cognitive Behavioral Therapy for Neuropathic Pain

Objective This evidence mapping is aimed at identifying, summarizing, and analyzing the available evidence on cognitive behavioral therapy (CBT) for neuropathic pain (NP). Methods This study was conducted following the methodology of Global Evidence Mapping (GEM). Searches were conducted in PubMed, Embase, the Cochrane Library, and PsycINFO to identify systematic reviews (SRs) with or without meta-analysis published before February 15, 2022. The authors independently assessed eligibility, extracted data, and evaluated the methodological quality of the included SRs using AMSTAR-2. The results were presented in the tables and a bubble plot based on the identified population-intervention-comparison-outcome (PICO) questions. Results A total of 34 SRs met the eligibility criteria. According to the AMSTAR-2, 2 SRs were rated “high,” 2 SRs were rated “moderate,” 6 SRs were rated “low,” and 24 SRs were rated “critically low.” The most common study design utilized to evaluate the efficacy of CBT for NP was the randomized controlled trial. In total, 24 PICOs were identified. Migraine was the most studied population. CBT for NP usually reaches the “potentially better” result at follow-up. Conclusions Evidence mapping is a useful way to present existing evidence. Currently, the existing evidence on CBT for NP is limited. Overall, the methodological quality of the included SRs was low. Further improvements in the methodological quality of SRs and more research on the most efficient CBT formats for NP are recommended in the future.


Introduction
Neurogenic pain (NP) occurs when the somatosensory nervous system is injured or diseased. It can be classified as peripheral or central neuropathic pain depending on where the injury or disease occurs [1,2]. It is a common chronic pain affecting 6.9% to 10% of the population, and more importantly, it has a significant impact on the quality of life [3]. The mechanisms and physiopathology of neuropathic pain are complex and varied. The potential causes include changes in ion channels, epigenetic regulation, and neuronal function, resulting in the abnormal activity of nerve fibers, peripheral and central sensitization, and so on [4,5]. Understanding the relationship between physiological mechanisms and clinical manifestations may help improve pain management. At present, the treatment of neuropathic pain includes pharmacological management and nonpharmacological management. Most guidelines for the management of neuropathic pain focus on pharmacotherapy, including anticonvulsants, tricyclic antidepressants, opioids, and drug combinations [6,7]. Since drugs may lead to adverse reactions and side effects, more and more studies are exploring the effects of nonpharmacological management of neuropathic pain, including psychotherapy, noninvasive neurostimulation therapy, and physiotherapy [8,9].
Cognitive behavioral therapy (CBT) is a prevailing psychological treatment for chronic pain. CBT was originally developed by Beck to treat people with depression. It is a structured, problem-oriented, and time-limited therapy [10]. As knowledge has evolved, there has been a third wave of cognitive behavioral therapy, which includes various techniques, such as acceptance and commitment therapy (ACT), mindfulness-based cognitive therapy (MBCT), and metacognitive therapy (MCT) [11,12]. CBT is a class of interventions that modify maladaptive cognitions to change emotional distress and maladaptive behaviors [11,13,14]. The effects of CBT in chronic pain may be related to structural changes in gray matter in brain regions associated with pain management and/or in the functional connectivity of these regions [15]. A study of chronic pain found that CBT significantly improved patients' clinical outcomes (treatment outcomes in pain survey, the chronic pain self-efficacy scale, and pain catastrophizing), and that the intrinsic functional connectivity (iFC) between the anterior default mode network and the amygdala/periaqueductal gray was reduced, while iFC between the basal ganglia network and the right secondary somatosensory cortex was increased compared to pain education [16]. In addition, one study found that after CBT intervention, patients had an increase in brain gray matter volume in the dorsolateral prefrontal and sensorimotor cortexes, which was associated with decreased pain catastrophizing [17]. CBT has been used to relieve neuropathic pain associated with spinal cord injury [18,19], diabetic peripheral neuropathy [20,21], and HIV-related peripheral neuropathy [22,23]. These randomized controlled trials have shown that CBT is feasible and has beneficial effects in patients with neuropathic pain. CBT is not a single intervention, and the programs used in these studies included relaxation, breathing techniques, cognitive restructuring, sleep hygiene, behavioral activation, mindfulness, and values clarification [18,20,22]. Although there are some differences in specific treatment techniques for different diseases, they all have important similarities. The various techniques are consistent by changing cognitions to reduce maladaptive behaviors [11,13,24]. Unfortunately, some studies had small sample sizes or did not have enough confidence in the results. An online cognitive behavioral pain management program has shown to improve worst pain intensity in patients with chemotherapy-induced peripheral neuropathy (CIPN), but the study had a small sample size and a high dropout rate [25]. A study on neuropathic pain after spinal cord injury showed that the pain intensity decreased after CBT, but no intervention effects (time * group interaction) were found, and the power analysis of pain intensity reduction was much lower than expected [19]. A study showed that CBT significantly reduced HIVrelated peripheral neuropathic pain, but 57% of the participants did not complete cognitive behavior intervention sessions [23].
Therefore, evidence of the effect of CBT on NP should be further explored and analyzed. Since NP is a clinical description but not a diagnosis and CBT is a class of interventions, systematic reviews (SRs) may not be the most effective in providing the comprehensive summary. This is because SRs are more focused on a specific clinical problem (specific intervention or exposure, or specific disease type). To cover a broad field, evidence mapping has become an emerging tool used to comprehensively identify and summarize current evidence on a topic [26][27][28][29]. Therefore, this study used evidence mapping to identify, summarize, and analyze the available evidence on CBT for NP, thereby assisting the clinical decision-making process.

Materials and Methods
We performed the evidence mapping based on the methodology proposed by the Global Evidence Mapping (GEM) [29] and previous articles [30][31][32]. The process was divided into four stages (see Figure 1).

Setting the Boundaries and
Context of the Map. To construct the evidence map, we reviewed research and guidelines related to CBT and NP, and consulted an expert in pain management. Afterwards, we established the criteria for considering studies according to the PICO: type of study: only SRs (with or without meta-analysis) were included because they provide the most reliable evidence; participants: the samples in the studies should be patients with NP; interventions: the interventions should be CBT; comparisons: the comparisons could be CBT, other treatments,    15,2022. Mesh terms and free-text terms were used for searching, such as "neuropathic pain", "neuralgia", "somatosensory disorders", and "peripheral nervous system diseases". In addition, the cited references of included studies were searched to ensure a comprehensive search. The detailed search strategies are available in Supplementary Material 1.
The search results were managed by EndNote X9. After removing duplicates, two reviewers (Li and Ding) screened the titles and abstracts independently and reviewed the full text of relevant literature to make a final decision. Any disagreement between the reviewers was resolved by consensus after consultation and discussion with an independent third reviewer (Zhu).

Methodological Quality Evaluation and Data
Extraction and Analysis. The methodological quality of all SRs included in the study was assessed using AMSTAR-2 [33]. AMSTAR-2 contains 16 items, of which items 2, 4, 7, 9, 11, 13, and 15 are the key items. The overall quality of SRs are divided into four levels: "high" means no or one nonkey item missing, "moderate" means more than one nonkey item missing, "low" means one key item with or without nonkey items missing, and "critically low" means more than one key item with or without nonkey items missing.
Then, the following two categories of data were extracted from the SRs: (a) General characteristics from SRs: authors, publication year, type of SR (with or without meta-analysis), aim, search date, design and number of included primary studies, and number of included participants (b) Characteristics of research questions: the PICO framework was described in detail according to the content of the included SRs. PICO contains four main contents: population (disease diagnosis), intervention (type, intensity, time, and frequency), comparison (type, intensity, time, and frequency), and outcome (impact of intervention) According to previous studies [30][31][32], the conclusions reported by authors were divided into five categories: "potentially better" (the conclusions reported a beneficial effect of CBT), "mixed results" (the conclusions of SRs with similar content were inconsistent), "no difference" (there was no significant difference between the intervention and comparison groups), "potentially worse" (the conclusions reported a less beneficial effect in the intervention group), and "unclear" (the conclusions were inconclusive or not reported by the authors) Two authors (Li and Hou) independently assessed the quality of the methodology and extracted the data. Any disagreement was resolved by consensus after consultation and discussion with an independent third reviewer (Zhu).      The x-axis represented the classification of the author's conclusions ("potentially better," "mixed," "no difference," "potentially worse," and "unclear").

Results
The selection of the eligible systematic reviews is shown in Figure 2. The list of the 88 excluded studies and reasons for exclusion are available in Supplementary Material 2.
3.1. Characteristics of Systematic Reviews. The characteristics of included SRs are shown in Table 1. We included 34 systematic reviews, 23 of which included a meta-analysis. The earliest review included was published in 2006, and more than half of the reviews were published in the last five years. Item 3  Item 4  Item 5  Item 6  Item 7  Item 8  Item 9  Item 10  Item 11  Item 12 Davies   Figure 3: Methodological quality of the included systematic reviews.    10 Neural Plasticity
3.3. Characteristics of PICOs from SRs. After merging the repeated studies extracted from SRs, 24 PICOs were finally extracted. The main characteristics are shown in Table 2. The detailed characteristics of PICOs before merging have been listed in Supplementary Material 3. For the population, 14 PICOs focused on migraines, which were the most extracted PICO questions. For the intervention, 15 PICOs used traditional CBT, 3 PICOs used ACT, 3 PICOs used cognitive restructuring or cognitive therapy, 2 PICOs used MBCT, and 1 PICO used mindfulness. In terms of the delivery type of psychological therapy, 6 PICOs used Internetbased CBT. In addition, PICOs had many forms of comparison, including waiting list/treatment as usual (10 PICOs), placebo/sham control (5 PICOs), education (4 PICOs), pharmacotherapy (2 PICOs), self-administered group (1 PICOs), relaxation (1 PICOs), and standard supportive psychotherapy (1 PICOs).
The evidence mapping of the CBT for NP is shown in Figure 4 by visualizing the data in Table 2. The evidence mapping found that the majority of CBT treatments showed "potentially better" or "mixed" results. Among them, CBT seems to have a better effect on patients with neuropathic pain after SCI and burning mouth syndrome (BMS). 12 PICOs included 20 primary studies reported as "potentially better," two of which involved 2 primary studies that were included in the same high-quality systematic review [54]. 10 PICOs were reported as mixed results due to inconsistent results, demonstrating the limited credibility of the results of these interventions. 10 PICOs included 12 primary studies that showed no difference in CBT compared with the control group. 3 PICOs included 3 primary studies which were rated as "unclear" due to unreported conclusions in the SR. There did not appear to be additional benefit from CBT compared to other types of psychotherapy, such as standard supportive psychotherapy or relaxation. 6 PICOs used Internet-based CBT, but the results of this delivery type were inconsistent. However, it is important to note that most of the studies which had reported the effects of follow-up had concluded that long-term follow-up was potentially beneficial.

Discussion
There are several ways to synthesize evidence and evaluate the effectiveness of interventions. We used an emerging approach called evidence mapping to visually display information in a user-friendly format [28]. This evidence map reviews studies published up to January 2022 and may be the first evidence map describing CBT for NP. Current research indicates a growing interest in the nonpharmacological treatments for neuropathic pain.
According to the analyzed results from the evidence mapping in this study, only a few studies are focusing alone on CBT for NP. Most SRs included a small number of original studies which focused solely on NP. Therefore, current evidence regarding CBT for NP may be limited. But the studies included were almost always randomized controlled trials, which are the best study design for evaluating the efficacy of treatments [92]. It is important to note that some of these studies were published earlier and had smaller sample sizes. Therefore, large-scale randomized controlled trials are necessary in the future. In addition, no PICO was identified as "potentially worse," and we cannot rule out the possibility that RCTs with negative conclusions are rarely published [93].
According to our results, the AMSTAR-2 scores of the SRs were generally low. To improve the methodological quality of future SRs, reviewers should pay more attention to the following aspects: to develop and provide an accessible protocol before conducting a review, to provide a list of excluded studies and the reasons for each exclusion, to discuss and explain the impact of risk bias on the results, to analyze and discuss the cause of heterogeneity and its impact on research results, and to report the sources of funding for included studies. Although evidence mapping does not require methodological quality assessment, some research results have suggested that the process is necessary to assess the reliability of conclusions [26,30].
CBT is not just an effective intervention for people with mental illnesses. According to our analysis, 34.29% of PICOs revealed that CBT is "potentially beneficial" to NP. At the same time, a number of studies reported conclusions with "mixed" (28.57% PICOs) or "no difference" (28.57% PICOs). 14 Neural Plasticity There were some differences in the conclusions of the current studies, and even the studies with "potentially beneficial" conclusions did not show large effect sizes. It may be due to the difference in CBT techniques, intervention duration, and intervention frequency. However, no study compared the effect of different CBT techniques on NP. Therefore, in the future, while updating the mixed conclusions, reviewers should go further in the efficacy of different techniques, treatment prescriptions, levels of supervision, and many other aspects of CBT for NP to seek better treatment strategies. In addition, we found that CBT was also effective on NP during follow-up, although these studies had a small to moderate effect size [61,62,64]. This indicated that patients would take more time to change their cognition and behavior. Differing from pain education, cognitive behavioral therapy is a structured approach. Patients may experience changes in ideas, cognitive processes, pain management behavior, and so on [10,94]. Therefore, long-term follow-up should be considered when optimizing trial programs in the future. Telemedicine has become the new way to provide healthcare with technological advancements. We noticed that 5 SRs [41,44,45,50,53] had evaluated the effectiveness of cognitive behavioral therapy (delivered remotely) for the management of chronic pain. The results of the studies showed that Internet-based CBT produced small but significant improvements. Although there is currently insufficient evidence to draw strong conclusions about the efficacy of Internet-based CBT for NP, these results are promising and can encourage future research in this field. There is no doubt that Internetbased psychological therapies offer many advantages, including addressing distance barriers and reducing health service costs. In the face of the COVID-19 epidemic, healthcare providers are also changing the way of pain management. Therefore, further research is warranted to increase confidence in using Internet-delivered CBT.

Limitations
This evidence mapping has some limitations. First, although we conducted our literature search in 2022, the included SRs conducted their literature searches in 2021 or earlier. Therefore, we could not investigate studies published recently but have not yet been included in the existing systematic reviews. Nevertheless, we do not believe this limitation would seriously affect the results of evidence mapping. Secondly, we used AMSTAR-2 to assess the methodological quality of the SRs, but not the GRADE to evaluate the evidence level, since our main study objective is finding research gaps to guide further research rather than supporting clinical decision-making. Thirdly, the conclusions of some SRs may be biased due to methodological limitations. However, we provided detailed reports for readers to evaluate each conclusion. Finally, we only included the SRs published in English because of the language barrier.

Conclusions
There is limited evidence about CBT for NP. Overall, the methodological quality of the included SRs was low. Evi-dence mapping is a useful method to guide further research to fill in current knowledge gaps. Existing research indicated that remote pain intervention could be effective. In the future, more investigation is needed to explore the optimal CBT formats for NP. Furthermore, the methodological quality of reporting SRs should be improved.

Data Availability
The data that support the findings of this study are from the published literature.

Conflicts of Interest
The authors have no conflicts of interest to declare.