Bibliometric and Visual Analysis of the Scientific Literature on Percutaneous Electrical Nerve Stimulation (PENS) for Pain Treatment

Abstract

Background: Percutaneous electrical nerve stimulation (PENS) is a minimally invasive peripheral neuromodulation approach implemented against chronic neuropathic and mixed pain. This bibliometric study aims to quantitatively evaluate the output of PENS for pain treatment in the scientific literature. The main purpose is to stimulate research in the field and bridge potential scientific gaps. Methods: Articles were retrieved from the Web of Science (WOS) database. The search key term was “percutaneous electrical nerve stimulation (All Fields) and pain (All Fields)”. Year of publication, journal metrics (impact factor and quartile, Q), title, document type, topic, and citations were extracted. The join-point regression was implemented to assess differences in time points for the publication output. The software tool VOSviewer (version 1.6.17) was used for the visual analysis. Results: One thousand three hundred and eighteen articles were included in the knowledge visualization process. A linear upward trend for annual new publications was found. Almost two-thirds of the documents were published in top-ranked journals (Q1 and Q2). The topic “efficacy” was prevalent (12.81%). Concerning article type, the search strategy yielded 307 clinical investigations (23.3%). Articles were cited 36,610 times with a mean of 42.4 citations per article. Approximately one-half of the articles were cited less than 23 times in a range of 21 years. The semantic network analysis for keywords found eight clusters. The analysis of collaborative efforts among researchers showed five thematic clusters including 102 authors with a minimum of five documents produced in collaborations. Most partnerships involved the United States, England, and Germany. Conclusions: despite the upward trend in the number of publications on the subject and the publication of articles in top-ranked journals, there is a need to increase scientific collaborations between researchers and institutions from different countries.

1. Introduction

Percutaneous electrical nerve stimulation (PENS) is a nonpharmacological, minimally invasive peripheral neuromodulation procedure. It is used, alone or as a complementary therapy, for addressing chronic neuropathic [1,2] and mixed nociceptive with neuropathic pain, especially in musculoskeletal disorders unresponsive to common therapies [3]. Clinical applications concern low back pain (LBP) [4], migraine and difficult-to-treat primary headache disorders [5], post-herpetic neuralgia [2], and scar tissue pain [2]. Moreover, PENS has also been used to address opioid-resistant cancer pain [6] and other unspecified chronic pain conditions [7].

This technique combines the principles of both electroacupuncture and transcutaneous electrical nerve stimulation (TENS) [1]. In particular, PENS works through a dedicated probe consisting of two or more needle electrodes (10, 32 gauge) inserted at the subcutaneous level and capable of conveying a mild electric current to a predetermined target (peripheral sensory nerve, soft tissue) with a dermatomal approach. The treatment plan includes one or more thirty-minute (15–60 min) sessions. The neurophysiological bases of the PENS functioning probably involve the modulation of the large-diameter afferent Aβ fibers and the local release of biochemical agents such as neurotransmitters and endorphins [3]. The stimulation of interneurons located in the Rexed laminae 2 and 3 of the dorsal horn was also suggested [8]. Furthermore, PENS can stimulate serotonin, cholinergic, and opioid receptors in the spinal cord, and ameliorate primary hyperalgesia through GABA signaling activation [3,8].

The scientific evidence on TENS is widely available [9,10]. On the contrary, despite the potential applications of PENS in different clinical conditions, there is no solid evidence to sustain this procedure. To date, evidence is mostly sustained by small-sized, controlled trials. In 2013, the National Institute for Health and Care Excellence (NICE) published guidance confirming the short-term safety and efficacy profile of the PENS technique. Furthermore, it encouraged the implementation of clinical research for establishing the selection of patients suitable for treatment and assessing long-term outcomes. The NICE document also stressed the need for clear documentation useful for the indications of treatment [2]. Previously, the American Academy of Neurology, the American Association of Neuromuscular and Electrodiagnostic Medicine, and the American Academy of Physical Medicine and Rehabilitation published guidelines and recommended PENS therapy for the treatment of painful diabetic neuropathy (level B) [11]. Thus, there is a need for an update.

Meta-analysis and bibliometric studies are both quantitative strategies. Nevertheless, meta-analyses are commonly implemented as theory extension instruments and summarize empirical evidence by analyzing the features and strength of outcomes and links encompassed by variables. On the contrary, bibliometric analysis summarizes the bibliometric architecture of a field by analyzing “structural relationships” between different research elements and features. In other words, bibliometrics is a computable approach that is suitable to investigate emerging trends and the insight structure within a given topic or field for gathering estimable, reproducible, and objective data. Therefore, this method can better broadly outline the knowledge of the research area. This approach is also useful for stimulating multidisciplinary collaboration [12].

On these premises, we conducted a visualized bibliometric network of all the literature available on the subject for highlighting the strengths and weaknesses of the scientific output.

2. Materials and Methods

2.1. Search Strategy and Data Collection

We adopted a strategy previously implemented in other bibliometric analyses [13,14]. The global literature on PENS and pain was scanned in the Web of Science (WOS) online database. The string used to find the corresponding manuscripts was “percutaneous electrical nerve stimulation (All Fields) and pain (All Fields)”. No language restrictions were inserted. All article types were included. The study period was from 1990 to 2022. All data were obtained on 26 April 2022. Findings were exported as a Microsoft Excel (.xlsx) file.

Titles and abstracts were independently screened by three authors (A.C., D.S., and D.N.). These independent researchers reviewed full-text versions of the articles and no pertinent papers were excluded. Crosschecking was performed, and any disagreements were resolved after discussion with the first and last authors (F.M. and M.C.). For manuscripts that met the requirements, we extracted data, including title, year of publication, journal, journal’s metrics as impact factors, and quartile (Q). The journal’s quartile is the category ranking and ranges from Q1 (best quartile) to Q4 (the worst quartile). The best Q was adopted for journals included in two or more categories. For journal metrics, the source was Journal Citation Reports™ 2021 (Clarivate Analytics). Other considered variables were document type, and the number of citations.

Furthermore, different topics were selected and extracted. They included “post-surgery scar tissue pain”, “post operative pain”, “tens”, “postherpetic pain cancer pain”, “polyneuropathy”, “complex regional pain syndrome (CRPS)”, “phantom pain”, “low back pain (LBP)”, “chronic musculoskeletal pain”, “shoulder pain”, “occipital neuralgia”, “fibromyalgia”, “myofascial pain”, “migraine”, “technique(s)”, “mechanism(s) of action”, “efficacy”, and “other”. The topics were chosen after careful evaluation of the titles of the articles by the study managers (F.M. and M.C.) in order to include aspects of preclinical and clinical research, application methodologies, and the potential fields of use of the techniques.

2.2. Analysis of the Annual Trend and Other Analyses

The trend in the annual number of the selected papers was computed by using join-point analysis (Joinpoint Regression Program software, version 4.9.1.0—April 2022; Statistical Methodology and Applications Branch, Surveillance Research Program, National Cancer Institute) and the Dickey–Fuller test. According to this approach, the average annual percent change (APC) in publication amount estimates the weight of the time variation. The Dickey–Fuller test is a commonly adopted unit root test. According to this approach, the null hypothesis is (1 − φ) = 0 versus the alternative (1 − φ) < 0 (where φ is the first-order lag coefficient of an AR model yt = a + φyt − 1 + εt) and hence large negative values of the test statistic led to the rejection of the null. The join-point regression (change point regression or segmented regression) estimates that data can be split into subsets. For example, in a time series with two distinct trends, it can be found that when t ≤ t^⁎, the model has a precise intercept and slope; when t > t^⁎, the model would have a different intercept and slope. Thus, t^⁎ represents a join-point. APC is a possible characterization of trend in study-focus rates over time and for discrete-scale counts. Assumption requires that rates change at a constant percentage of the rate of the previous scale period. Rates that change at a constant percentage every year change linearly on a log scale. Thus, to calculate the APC for a series of values, the following regression model is implemented:

$${\displaystyle \ln \left({{r}}_{{t}}\right)={\alpha }+{\beta } {t}}$$

where r_t is the rate of the t^th year. The APC from that year to the next one is calculated as:

$${\displaystyle {A}{P}{C}\left({t},{t}+1\right)= \frac{{{r}}_{{t}+1}-{{r}}_{{t}}}{{{r}}_{{t}}} \cdot 100=\left({{e}}^{{\beta }}-1\right)\cdot 100}$$

The model focuses on linear segments connected at join points that express the optimum fit of the observed data. These elements reduce the amount of the square of the differences between estimated and observed data. Consequently, the join-point year represents the point in time where the trend change is evaluated [15].

The suite ggplot2 of the R software, version 4.1.3 (R Core Teams, R Foundation for Statistical Computing, Vienna, Austria), was adopted for graphical visualization (quartiles, article types, topics, citations).

2.3. VOSviewer-Based Knowledge Visualization

For knowledge visualization, we used the software tool VOSviewer (version 1.6.17). It structures networks of linguistic elements mined from a set of scientific articles, develops clusters, and provides standard weight attributes including “Links” and “Total link strength” of the elements (articles, keywords, authors, countries, citations, and others). A graphical visualization is produced [16]. The following features were investigated [17]:

• Co-occurrence of keywords: also termed semantic network; studies plausible links between terms (interconnection).
• Co-authorship analysis for authors: expresses the collaborative efforts among researchers.
• Co-authorship for countries: shows international partnerships.

The flowchart of the study is shown in Figure 1.

3. Results

3.1. Bibliometric Analysis of Publication Output

A total of 1353 articles on PENS therapy for pain management were published from 1990 to April 2022 in WOS. After the exclusion of 35 papers, 1318 articles were included in the knowledge visualization process [18].

The annual trend of the retrieved articles is shown in Figure 2.

Over the whole study period (1990–2022), a linear upward trend for the number of annual new publications was observed (Dickey–Fuller test, p = 0.73). The annual trend investigation (join-point analysis) found two segments (time periods) with different slopes. In the interval 1990–2020, there was a significant increment in the annual percent change (APC) (APC = 6.05: 95%CI 7.4–9.5; * p < 0. 01). Concerning the 2021–2022 (April 2022) period, the curve was not significant (APC = −35.1; p = 0.34). (Figure 3).

An amount of 795 articles was published in Q-ranked journals (Q1 to Q4). The quartile analysis demonstrated that 45% of the documents (n = 358) were published in Q1 journals (Figure 4).

Regarding the most productive journals, the journals “Pain” (IF 7.926; Q1) and “Neuromodulation” (IF 3.025; Q3) published 52 and 49 articles, respectively (

Table 1. Metrics of the 10 most productive journals.

Journal	Impact Factor °	Best Quartile °	Articles (n)
Pain	7.926	Q1	52
Neuromodulation	3.025	Q3	49
Neuroscience Letters	3.197	Q3	18
Pain Physician	4.396	Q1	17
Regional Anesthesia and Pain Medicine	5.571	Q2	15
Neuroscience	3.708	Q3	15
Current Pain and Headache Reports	3.904	Q2	14
Pain Practice	3.079	Q2	14
Experimental Brain Research	2.064	Q4	11
European Journal of Pain	3.651	Q2	11

Legend: ° source—Journal Citation Reports™ 2021.

).

With respect to article types, 307 were clinical investigation articles (186 observational studies, 121 randomized controlled trials, RCTs), 229 were reviews (182 narrative reviews and 47 systematic review/metanalysis) and 53 were case reports. An amount of 131 documents concerned preclinical research (Figure 5).

The most relevant topics of the selected papers were “efficacy” (n = 212), “TENS” (n = 210), and “mechanisms of action” (n = 145) (Figure 6).

Articles were cited 36,610 times with a mean of 42.4 (SD 83.2) citations per article. Considering four different time intervals (1990–99; 2000–10; 2011–15; >2015), we found a solid heavy left tail for each period. In fact, the median of citations was 23 articles (IQR = [7, 50]) and one-half of the articles were cited less than 23 times in a range of 21.3 years (Figure 7).

The most cited article (1772 citations) was published by Rossini et al. [19] in 2015. It is a review of the theoretical, physiological, and practical aspects of different neuromodulation strategies released by the International Federation of Clinical Neurophysiology (IFCN). The top cited articles [19,20,21,22,23,24,25,26,27,28] are reported in

Table 2. Summary of the most cited articles. Abbreviations: Q—quartile.

Author(s)	Article Title	Journal	Quartile	Citations (n)	Year
Rossini PM, et al. [19]	“Non-invasive electrical and magnetic stimulation of the brain, spinal cord, roots and peripheral nerves: Basic principles and procedures for routine clinical and research application. An updated report from an IFCN Committee”	Clin Neurophysiol	Q2	1772	2015
Gracely RH, et al. [20]	“Painful neuropathy: altered central processing maintained dynamically by peripheral input”	Pain	Q1	635	1992
Yaksh TL, et al. [21]	“Current and Future Issues in the Development of Spinal Agents for the Management of Pain”	Curr Neuropharmacol	Q1	498	2017
Cruccu G, et al. [22]	“EFNS guidelines on neurostimulation therapy for neuropathic pain”	Eur J Neurol	Q1	420	2007
Tracey I, et al. [23]	“Imaging attentional modulation of pain in the periaqueductal gray in humans”	J Neurosci	Q1	391	2002
Olesen J, et al. [24]	“Origin of pain in migraine: evidence for peripheral sensitization”	Lancet Neurol	Q1	366	2009
Deer TR, et al. [25]	“Neuromodulation appropriateness Consensus Committee. The appropriate use of neurostimulation of the spinal cord and peripheral nervous system for the treatment of chronic pain and ischemic diseases: the Neuromodulation Appropriateness Consensus Committee”	Neuromodulation	Q3	347	2014
Cruccu G, et al. [26]	“The lower limb flexion reflex in humans”	Prog Neurobiol	Q1	339	2005
Deumens R, et al. [27]	“Repairing injured peripheral nerves: Bridging the gap”	Prog Neurobiol	Q1	339	2010
Sheffler LR, et al. [28]	“Neuromuscular electrical stimulation in neurorehabilitation”	Muscle Nerve	Q2	333	2007

.

3.2. Knowledge Visualization

The bibliometric analysis of the keywords was centered on the terms that occurred more than five times in the database. Of the 3582 keywords, 45 reached the threshold. Eight clusters were collected. The most used keywords were ELECTRICAL STIMULATION (occurrence 310; total link strength 302); PAIN (183/180), and NEUROPATHIC PAIN (173/170) (Figure 8).

The map of collaborative efforts among researchers indicates the authors that cooperate in this research field. It was conducted with a cutoff of at least four documents for a single author and written in collaboration with other authors included in the dataset. Thus, of 5140 authors, 102 met the threshold and five clusters were found. In one cluster, Dr. Edward R. Mariano from Stanford University was the most representative author for co-authorships. In the other clusters, the software indicated Drs. Joseph W. Boggs, Brian M. Ilfeld, John Chae, and Alaa Abd-Elsayed. (Figure 9).

For the co-authorship analysis for countries, the minimum number of documents (threshold) per country was five. Therefore, of 61 countries, 31 met the threshold. The overall strength of the co-authorship links with other countries was calculated. Nine clusters were collected with 161 links and a total link strength of 540. Most partnerships involved the United States, England, and Germany. In the United States, 497 documents with 26 links were listed; England and Germany collected the same number of documents (n = 125) and links (n = 21) (Figure 10).

4. Discussion

A bibliometric analysis is a profitable opportunity for highlighting potential gaps in the literature. Consequently, it can offer valuable suggestions to direct the research [12]. To our knowledge, this is the first bibliometric analysis conducted on the applications of PENS therapy for pain management. Therefore, results are potentially useful for planning further research in the field. The results of the bibliometric investigations indicated an upward annual trend in the number of scientific output and the publication of articles in top-ranked journals. On the other hand, partnerships among researchers should be implemented.

The annual trend of scientific production on the subject is progressively growing. This finding suggests an interest in the technique and its applications. However, the comparative analysis conducted by time intervals (trend analysis) highlighted a significant increase in articles only up to the year 2020. On the contrary, from 1 January 2021 to April 2022, the increase remains, but it was not significant. These data need a necessary update to verify whether they express a trend reversal or are only a transitory phenomenon. For example, in the years of the pandemic, other topics may have ceased to be preferred by researchers.

Concerning the quality of the journals, two-thirds of the articles were published in top-ranked journals (Q1 plus Q2, 75%). Many articles were published in reference journals in the field of pain medicine with a high impact factor. For example, the journal “Pain” published 52 papers. The analysis of sources is helpful for authors to search the literature and choose a journal for the submission of their articles.

Overall, most of the papers (98%) reached less than 100 citations, although we calculated a high mean of citations per article (n = 42.4) and approximately one-half of the documents received an important number of citations for over 20 years. Although the citation trend should indicate the visibility of an article rather than its overall goodness [29], taken together, these findings are suggestive of the high quality of the research. Nevertheless, only a quarter of the articles concern clinical research and 44% of the articles are meeting abstracts, book chapters, corrections, and other non-research or -review articles. These data represent an element of weakness, as clinical studies are essential to obtaining the scientific evidence useful for recommendations and guidelines.

Approximately 10% of the articles concerned preclinical research and, in same the percentage, the topic “mechanism of action” was extracted. This demonstrates that there is interest in explaining the mechanisms underlying the analgesic effect of the technique. The precise processes of PENS-induced analgesia are not yet recognized [30]. Therefore, a better understanding of PENS’s actions will certainly stimulate translational research. For example, telecommunication engineering focused on the study communication between neurons and transmitting signals through nanomachines can offer important data and an appropriate research methodology [31,32].

The most relevant topic was “efficacy” (13.6%). Data on the actual efficacy of the treatment are of fundamental importance in order to plan further research. Recently, Beltran-Alacreu et al. [33] conducted a systematic review and metanalysis to determine if the use of PENS is more effective than TENS for treating musculoskeletal pain. They included nine RCTs in the qualitative analysis and seven in the quantitative analysis and found no difference between the two approaches in terms of pain intensity. Probably, when compared to TENS, PENS could induce a decrease in medication intake. Nevertheless, it is difficult to obtain definitive results due to important differences in protocols, parameters (e.g., frequency of pulsed current), and treatment plans. Other clinical trials were conducted for evaluating the efficacy in different clinical contexts. For example, in patients with myofascial chronic neck pain (n = 40) treated with dry needling (DN) combined with high or low-frequency PENS, Hernandez et al. [34] found significant improvements in the pain scores in both groups. Another RCT demonstrated that, in the short-term, PENS is more effective than DN in patients with mechanical neck pain [35]. The main limitation of these trials is the small sample size.

Keyword analysis is a profitable approach to defining preeminent themes and provides a framework of documents on a research field. The VOSviewer-based knowledge visualization showed an acceptable interconnection between the terms (keywords). This semantic network analysis found eight clusters, indicating a varied distribution of the keywords used and translates into a wide range of topics covered. On the contrary, knowledge-mapping suggested that scientific collaborations between researchers, or institutions, mostly concern the United States, England, and Germany. Furthermore, the co-authorship analysis of authors suggested that collaborations between authors are lacking. Remarkably, only five poorly structured clusters were collected. Since the cooperative networks show limited group characteristics, the authors tend to publish independently. Therefore, the development of research across different countries is unbalanced and partnerships should necessarily be strengthened. The reasons for these gaps need to be analyzed. Multicenter studies, for instance, are very difficult to carry out. Participating centers should be equipped with the same devices; inclusion and exclusion criteria should be adequately established, as well as operational protocols. It presupposes an important economic commitment. Rossi et al. [7] completed a multicenter national study in Italy. In their prospective observational investigations, the authors successfully treated patients affected by peripheral neuropathic pain, localized and refractory to pharmacological therapies. The same authors reported that the small sample size (n = 76), the non-randomized design, and the high prevalence of some disease conditions (post-herpetic and occipital neuralgias) affected the reliability of the results.

4.1. Limitations

This analysis has different limitations. The search and methods were limited to articles extracted and analyzed from the WoS database. Although it includes all SCI- and SSCI-listed journals, this strategy may have excluded articles present in other datasets (e.g., Scopus). Moreover, a citation search in WOS is not exhaustive (e.g., some subject areas are poorly covered). However, this approach is widely adopted for bibliometric research.

As Gasparyan et al. [36] recently suggested, the values of articles and metrics should be evaluated given a set of confounders including patterns of citation and variability related to study context and academic disciplines. The main purpose of our study was to present an overview of the topic to stimulate research and bridge scientific gaps. Detailed bibliometric and scientometric research will be needed to focus on the different indicators. Moreover, other software tools such as CiteSpace (which also removes duplicates or publications with an unclear year of publication) can be used for an in-depth visualization [37]. We adopted VOSviewer, as it focuses on visualizations at an aggregate level.

Other analyses such as the Price’s index [38], which is useful for accessing the trend in a specific country or a given discipline, as well as co-citation analysis (bibliographic coupling) were not performed. Since we presented a citation analysis (divided into four periods) and a descriptive analysis of the top-10-cited articles, the rate with which two articles are cited together by other documents is probably not useful for macroscopic description of the phenomenon. Supplementary aspects to be investigated were the impact of open access on the annual trend and quality of articles, the multidisciplinary background of researchers, and the involvement in collaborative networks of organizations.

The string we used for bibliometric analysis (“percutaneous electrical nerve stimulation (All Fields) and pain (All Fields)”) yielded a non-exhaustive number of articles with the exclusion of documents of particular interest. However, our analysis focused on the applications of PENS for pain management and other clinical use were intentionally excluded. Moreover, since the exported records from the WOS core collection included the “full record”, we tried to intercept the keyword “pain” in all articles. These limitations have also been mentioned in other bibliometric investigations [13,14,29,37,38].

Finally, the increase in publications for our research topic should be put in relation to the general increase of publications in WoS. It was estimated, for example, that the annual growth of publication production in the world is up 4% from 2010 to 2020 with a constant growth trend [39]. Consequently, our results concerning these aspects could be biased and probably not relevant at all because all publications increase during the same timeframe. Moreover, the last two years of the analysis must be carefully evaluated, because, due to publication procedures, not all articles being published in those years are already listed in WoS/accessible, this is especially true for 2022. We have chosen not to exclude the last few years a priori, to provide more extensive information in numerical terms. This implies that the analysis will have to be updated periodically.

4.2. Suggestions for Future Research

A bibliometric study is primarily intended to offer research suggestions. Outcome analysis is a key aspect of the research. In terms of co-occurrence, in addition to the word “efficacy”, our bibliometric study produced poor results in this direction. This finding reflects the limited data in the literature. Moisset et al. [5], for example, conducted a meta-analysis on neuromodulation techniques for acute and preventive migraine treatment. Regarding PENS, the authors found only one high-quality RCT with detailed primary (i.e., change in monthly migraine days and the 50% responder rate) and secondary (i.e., monthly migraine attacks, monthly headache days, and monthly acute antimigraine drug intake) outcomes [40].

The most important gap to be addressed is the standardization of methods and the definition of protocols for the various problems to be treated. In other words, the lack of uniformed application criteria limits the judgment on the effectiveness of the technique. Although PENS is apparently a simple intervention, practitioners must customize their treatment strategy based on several variables. The clinical definition of the problem and the indications (e.g., allodynia and other positive neuropathic features), the position of the electrodes, and the size of the area to be treated must be well established. In addition, electrical characteristics such as pulse width, frequency, pattern, and duration of treatment must be modulated according to pain (intensity characteristics, type, duration, etc.). Therefore, good practice guidelines are needed. In a comprehensive appraisal of the features and efficacy conducted on 169 reviews articles and 49 meta-analyses, the authors concluded that the methodological shortcomings of the RCTs hampered confident judgments on efficacy. For example, the authors underlined that the majority of studies had small and inadequate sample size as well as inconsistencies in methodologies, analyses, and findings. Finally, the absence of an analysis of adverse events is a serious gap [41].

5. Conclusions

This bibliometric analysis showed that the produced literature on the topic is progressively increasing. Notably, the articles are usually published in top-ranked journals and receive an important number of citations over the years (up to over 20 years from publication). On the other hand, there is a need to increase clinical research as well as scientific collaboration among researcher teams and institutions from different countries.