The dose-dependent effects of transcutaneous electrical nerve stimulation for pain relief in individuals with fibromyalgia: a systematic review and meta-analysis

Abstract

Transcutaneous electrical nerve stimulation (TENS) is a nonpharmacological modality widely used to manage pain; however, its effectiveness for individuals with fibromyalgia (FM) has been questioned. In previous studies and systematic reviews, variables related to dose of TENS application have not been considered. The objectives of this meta-analysis were (1) to determine the effect of TENS on pain in individuals with FM and (2) determine the dose-dependent effect of TENS dose parameters on pain relief in individuals with FM. We searched the PubMed, PEDro, Cochrane, and EMBASE databases for relevant manuscripts. Data were extracted from 11 of the 1575 studies. The quality of the studies was assessed using the PEDro scale and RoB-2 assessment. This meta-analysis was performed using a random-effects model that, when not considering the TENS dosage applied, showed that the treatment had no overall effect on pain (d+ = 0.51, P > 0.050, k = 14). However, the moderator analyses, which were performed assuming a mixed-effect model, revealed that 3 of the categorical variables were significantly associated with effect sizes: the number of sessions (P = 0.005), the frequency (P = 0.014), and the intensity (P = 0.047). The electrode placement was not significantly associated with any effect sizes. Thus, there is evidence that TENS can effectively reduce pain in individuals with FM when applied at high or at mixed frequencies, a high intensity, or in long-term interventions involving 10 or more sessions. This review protocol was registered at PROSPERO (CRD42021252113).

1. Introduction

Fibromyalgia (FM) is a chronic neurological health condition that causes widespread chronic muscle pain, fatigue, and tenderness with hyperalgesia to pressure over tender points and is usually accompanied by sleep, cognitive, and mood disorders.^15,98,99 Fibromyalgia has a prevalence of between 0.2% and 6.6% and affects 6.6% of the population in the United States.⁵⁸ It is more common in women, with a female-to-male ratio of 3:1, and its incidence also increases from 30 years onwards.⁷⁰ Although FM etiology is unknown, the data suggest that ineffective descending pain inhibition^50,85,89 and enhanced excitability of the central nervous system^69,87,89 may result in abnormal stimulus processing, leading to a general increase in pain perception^86,92 that may last for many years and can be disabling, reducing both the patients' quality of life and their emotional well-being.^15,41,76,94

The available treatments do not have a curative intent, but rather aim to provide symptomatic relief to improve the patients' physical condition and quality of life.^56,67,69 The updated guidelines published by the European League Against Rheumatism (EULAR, 2017) for FM management recommend first focusing on nonpharmacological therapies.⁵⁶ There is some evidence that exercise is an effective treatment,⁵ and it is thus the only highly recommended therapy–based intervention for this disease.⁵⁶ However, the pain experienced can limits the ability to participate in physical activities and interfere with subjects' adhering to exercise plans.^24,88 The use of nonpharmacological treatments to reduce pain would therefore in theory enable greater activity and thus notably improve their quality of life.^3,61,91

A potential nonpharmacological pain control treatment in individuals with FM is transcutaneous electrical nerve stimulation (TENS). Transcutaneous electrical nerve stimulation delivers pulsed electrical currents across the intact surface of the skin through conducting electrodes. This stimulates the peripheral nerves and elicits physiological nervous system responses, usually resulting in pain relief. Transcutaneous electrical nerve stimulation has been primarily used for pain relief irrespective of its origin, duration, or setting (ie, inpatient, outpatient, or palliative care).^38,40 In addition, its low cost, portability, and ease of autonomous use by the patient, as well as the absence of significant adverse effects compared with pharmacological treatments, could make TENS a viable treatment option for chronic pain conditions like FM.^41,93

Although randomized controlled trials have shown effectiveness of TENS for several chronic pain conditions, systematic reviews have shown mixed results.²⁹ The consensus on the reasons for the inconclusive results regarding TENS in reviews and meta-analyses seems to be both a lack of well-designed, randomized, controlled trials (RCTs) and difficulty in defining accurate TENS treatment output parameters.^29,41,93 The latter include the key parameters for TENS dosing: (1) the frequency of the electric pulses (hertz)^44,78,80; (2) intensity (milliamperes) of the stimulation^{1,4,6,35,41,65}; (3) electrode application sites³⁷; and (4) the number of sessions.⁸³ Variation in these parameters causes different physiological effects or can cause the absence of any therapeutic effect,^23,43,97 whereas variation in other TENS parameters (waveforms, pulse durations, pulse patterns) has not been demonstrated to have significant hypoalgesic effects on clinical outcome.³⁹ Bennett et al.⁴ demonstrated that suboptimal dosing was particularly prevalent in RCTs about TENS. Thus, trials with ineffective TENS dosages may have contributed to the negative outcomes in previous meta-analysis studies on chronic pain.⁶

The evidence for the effectiveness of TENS on FM is also conflicting. Thus, 2 systematic reviews^26,41 and 1 review with a meta-analysis⁷⁴ concluded that there was insufficient evidence for the effectiveness of TENS for pain relief in individuals with FM. However, none of the reviews of TENS on FM we consulted had analyzed the potential different dose-dependent effects of TENS in detail. Thus, we undertook this meta-analysis of the clinical efficacy of TENS in individuals with FM by evaluating pain outcomes from a dose–response perspective. To the best of our knowledge, this is the first time that analysis of this type of TENS on FM has been conducted. The primary objective of this review was to determine the effect of TENS on pain in individuals with FM. Our secondary objective was to determine the effect of different TENS dose parameters (frequency, intensity, application site, and number of sessions) on pain reduction in individuals with FM.

2. Methods

This systematic review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The protocol for this work was prospectively registered on the Prospective Register of Systematic Reviews (PROSPERO) on May 26, 2021 (CRD42021252113) and the entirety of the text is available online (www.crd.york.ac.uk/PROSPERO/display_record.php?RecordID=252113).

2.1. Search strategy

The MEDLINE (through PubMed), Cochrane Central Register of Controlled Trials, Physiotherapy Evidence Database (PEDro), and EMBASE electronic databases were searched from their start until October 1, 2022. Reference lists of potentially eligible studies and systematic reviews were manually screened and were used as additional sources to ensure that every relevant study would be identified. Search terms included MeSH headings and synonyms of keywords associated with the population (individuals with FM) and intervention (transcutaneous electrical nerve stimulation). The search was limited to studies on humans and articles published in English, Spanish, Portuguese, or French. The search terms and strategies we used are described in supplementary file 1 (S1) (available as supplemental digital content at https://links.lww.com/PAIN/B786).

2.2. Eligibility criteria and selection

In this review, trials were included if (1) the design was a randomized controlled trial; (2) the participants were adults (aged 18-70 years) diagnosed with FM using any recognized diagnostic criteria; (3) TENS was applied as the only therapy or in combination with other treatment types; (4) the outcome measure was a reported change in pain intensity from the baseline levels to the follow-up, measured using a visual analogue scale or a numeric rating scale; and (5) TENS was compared with a nonintervention control, placebo control, or other pain relief intervention such as pharmacological treatment or another nonpharmacological treatment (hydrotherapy, heat, acupuncture, or dry needling, etc). We excluded studies performed with any other forms of electrical stimulation, including invasive techniques such as those involving inserting needles or electrodes into the skin (eg, electroacupuncture or percutaneous electrical nerve stimulation) or electrical stimulation that does not elicit any feelings (microcurrents).

All the articles identified in any of the databases using the search strategy described above were combined, and duplicates were excluded. The titles and abstracts of all these articles were independently evaluated by 2 researchers (J.J.A. and J.F.L). Studies that did not meet the eligibility criteria according to the titles or abstracts were excluded. Any abstracts that provided insufficient information concerning the inclusion and exclusion criteria were selected for full-text evaluation. Disagreements regarding study the eligibility were discussed between the 2 reviewers (J.J.A. and J.F.L.), and if there was no consensus, a third reviewer (M.D.A.) reviewed the text.

2.3. Data extraction

We developed a standardized form for data extraction in this review. The data extracted included the author, study title, objective, randomization methods, treatment allocation, blinding, comparison groups, participant characteristics, number of sessions, TENS parameters, TENS stimulation sites, adverse effects, and baseline and postintervention pain outcomes. Given that most authors assessed the pain levels at rest, in the case that pain was reported both at rest and in movement, we extracted the data for the resting values. Moreover, most studies assessed pain immediately after treatment; therefore, if more than one pain outcome was reported over time, we took the data closest to the time of the treatment for analysis.

To conduct a subgroup analysis of the trials based on the dose-dependent effects of TENS, 4 TENS parameters in each trial were scored as being an “appropriate” or “inappropriate” delivery of TENS stimulation. We considered the appropriateness of the TENS treatment administration according to each of the 4 following parameters.

• (1) Delivery at a frequency greater than 10 Hz to 200 Hz (a high frequency) because this was more likely to be comfortable and effective in chronic pain than low-frequency TENS therapy (10 Hz or less).^{10,39,41,44,51,52,79,83} Mixed frequencies (changing between low and high frequencies within the same treatment session) were also considered appropriate. This is because this delivery modality may overcome the analgesic tolerance that can occur when TENS is applied as a long-term repeated treatment.^10,52
• (2) Use of a TENS intensity able to elicit a feeling described as between “strong but comfortable” and “the highest tolerable nonpainful” level. This is because high intensities are associated with significant reductions in pain compared with lower intensities, which may be ineffective.^{1,13,15,17,49,62,68}
• (3) Electrode placement at the site of the pain reported by the patient or on the nerve bundles near the site of the reported pain.^39,42 This was because the same electrode placement would not be considered optimal for all the participants because the placement had to fit the specific pain experience of each patient.
• (4) At least 10 sessions completed given that TENS has been proven to produce a cumulative effect in a variety of chronic pain conditions.^11,25,48,57

In the case that any of the outcome data were unclear or not reported, we contacted the authors for further details. Two researchers (J.J.A. and J.F.L.) independently performed data extraction for each of the included studies. Disagreements were resolved through discussion between the 2 reviewers and, if required, the opinion of a third reviewer (M.D.A.) was requested.

2.4. Assessment of methodological quality and bias risk

The methodological quality and risk of bias assessment of each of the included studies was independently performed by 2 reviewers (J.J.A. and L.S.M.) by applying the PEDro scale²¹ and version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2).³² PEDro is an 11-item scale used to assess the risk of bias in the following: eligibility criteria, random allocation, concealed allocation, similarity at baseline, participant blinding, therapist blinding, assessor blinding, >85% participant retention, intention-to-treat analysis, reported between-group difference, point measures, and measures of variability. The validity of the PEDro scale to measure methodological quality in clinical trials has been proven.²¹ Items 2 to 11 on the scale contribute to the internal validity, and each fulfilled item contributes 1 point to the total score (ranging from 0 to 10). The first item, which relates to external validity, is not scored. Studies that obtained scores greater than or equal to 6 of 10 on the PEDro scale were considered to be of a moderate to high quality. Disagreements were resolved through discussion by the 2 reviewers.

The RoB 2 tool³² was used to assess the internal validity of the studies. The domains assessed for risk of bias were sequence generation, allocation concealment, participant blinding, personnel and outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias. For each domain, a judgment was assigned with a “yes” answer to indicate a minimal risk of bias, a “no” indicating an elevated risk of bias, or “unclear” indicating an unclear or unknown risk of bias where studies had reported insufficient details.

2.5. Data synthesis

The effect size index was the standardized mean difference between the pretreatment and posttreatment change in scores in the treatment and control groups.⁶³ This index was computed, for each comparison between treatment and control groups and outcome (k), by subtracting the mean pretest–posttest difference of the control group (M_CPre and M_CPost) from the mean pretest–posttest difference of the TENS group (M_TPre and M_TPost) and then dividing this difference by the pooled pretest standard deviation of both groups. Therefore, the means and standard deviations of the outcomes at baseline and immediately after the intervention and the number of participants were extracted for this work. Because d has a slight positive bias, that is, a tendency to overestimate the value of the effect size in small samples, Hedge g [g = c(m) × d] was applied as an unbiased estimate, where c(m) = 1 to 3/(4N − 9) is a correction factor for the d index for small sample sizes, and N is the total study sample size.³¹ Positive d values indicated a better result in the TENS group compared with the control group and vice versa. Effect sizes of 0.20, 0.50, and 0.80 referred to small, moderate, and large effect sizes, respectively.¹⁷

In the RCTs with 2 treatment groups and 1 control group, such as that of Lauretti et al.,⁴⁶ the sample size of the control group was divided by 2 to be able to make 2 comparisons while avoiding statistical dependence. In the same say, in the RCTs with 1 treatment group and 2 control groups, such as those of Dailey et al.,^18,20 the sample size of the treatment group was divided by 2 to be able to make 2 comparisons while avoiding statistical dependence. To assess the reliability of the effect size calculations, 2 researchers independently computed the effect size (double effect size calculation process). Interrater reliability was satisfactory, with a mean intraclass correlation of 0.976 (SD = 0.034), ranging from 0.952 to 1.

2.6. Statistical analysis

The pooled standardized mean change difference (d₊) and its corresponding 95% confidence interval (CI) were calculated assuming a random-effects model in the statistical calculations. In the absence of heterogeneity, the results yield of this model were identical to those of a fixed-effects model, and they allowed the conclusions to be generalized to a wider array of situations.⁷ In addition, the statistical significance of the pooled standardized mean difference was assessed using the Z test. A forest plot was constructed to represent the individual and pooled effect size estimates with their 95% confidence intervals (CIs) and to allow visual inspection of the study heterogeneity. In addition, both Cochran Q-statistic and the I² index were calculated to assess the heterogeneity of the effect sizes.³³ A Q-statistic with P < 0.05 was indicative of heterogeneity among the effect sizes, and the degree of this heterogeneity was estimated using the I² index. Thus, I² values of around 25%, 50%, or 75% denoted a low, moderate, or large amount of heterogeneity, respectively.³⁴

We used variability moderator analyses to examine the influence of moderator variables on the effect size. Weighted analyses of variance (ANOVAs), assuming a mixed‐effects model, were applied for categorical moderators by the improved method proposed by Knapp and Hartung⁴⁵ and as also described in Viechtbauer et al.⁹⁶ In addition, an estimate of the proportion of variance accounted for by the moderator variable was computed following Raudenbush's (2009) proposal.^55,72 The model misspecification was assessed with the Q_W statistic.

Finally, given that studies included in this meta-analysis were previously published, we conducted analyses to determine whether publication bias might be a threat to the validity of the results of the meta-analysis. Thus, Egger test was a simple unweighted regression, which used the precision of each study as the independent variable (defining precision as the inverse of the standard error of each effect size) and the effect size divided by its standard error as the dependent variable. A nonstatistically significant Student t-test result for the hypothesis of an intercept being equal to 0 meant that publication bias could be eliminated as a threat to the validity of the pooled effect.⁹⁰ All statistical analyses were conducted using the metafor package in R⁹⁵ and were all interpreted assuming a significance level of 5% (P < 0.05) using 2-tailed tests.

3. Results

3.1. Characteristics of the included studies

The initial electronic database search identified 1575 potentially relevant trials. An additional 2 trials were found through reference list screening. After removing duplicates, 486 trials were screened for eligibility based on the title and abstract. Forty-seven articles were assessed for full-text eligibility; 11 of these met the inclusion criteria and were included in this review (Fig. 1).

Figure 1.: PRISMA (2020) flow diagram. PRISMA, preferred reporting items for systematic reviews and meta-analyses.

A total of 789 participants were included in all the considered studies; of whom, 392 received TENS alone or as a combination with another treatment. The mean age of the participants ranged from 30.0 to 52.9 years, and most of them were women (the mean number of females ranged from 75.7 to 100). The higher disease prevalence was among those aged 30 years,⁷⁰ and the 3:1 ratio matches the data already established for this population in the literature.⁷⁰ In 9 of the 11 studies, pain was measured using a visual analogue scale,^{8,9,18,20,46,54,67,80,95} which have been widely used to assess pain intensity and whose validity and reliability has been shown in many studies, whereas 2 of the 11 studies used other numerical scales, the Myalgic Pain Score⁶⁵ and the Fibromyalgia Impact Questionnaire Pain Item.³⁶ The average baseline pain intensity reported ranged from 4.62 to 8.14 on a 10-point scale, which corresponded to a level greater than mild pain. This prevented the appearance of a “floor effect” associated with insufficient baseline pain that could potentially be relieved post intervention and is representative of the clinical situation of the population experiencing FM. No between-group differences at baseline found in any of the trials included (Table 1). The studies were also homogeneous regarding the diagnostic criteria used to include individuals with FM because the majority (8 of the 11 studies) included used the same American College of Rheumatology (ACR) 1990 diagnostic criteria as inclusion criteria,^{8,18,20,46,54,64,67,80} whereas 2 studies used ACR 2010,^9,36 and 1 did not report which version of the ACR was used.¹⁰⁰

Table 1 - Characteristics of the included studies.

Author, year	Design	Interventions groups	Sample size per group	FM diagnostic criteria	Female (% sample)	Age ± SD	Outcome (ranges)	Pain rating while TENS on or immediately after	Baseline pain intensity ± SD
Castro-Sánchez et al., 20209	RCT	TENS/dry needling	37/37	ACR 2010	75.7/86.5	47.84 ± 8.12/49.35 ± 5.82	Visual analogue scale (0-10)	No	8.14 ± 1.27/7.86 ± 1.18
Carbonario et al., 20138	RCT	Exercise + TENS/exercise	14/14	ACR 1990	100/100	52.9 ± 5.9/51.9 ± 9	Visual analogue scale (0-10)	No	7.7 ± 2.7/7.8 ± 1.9
Da Silva et al., 200877	RCT	TENS/hydrotherapy	5/5	ACR 1990	100/80	50.6 ± 13.4/47.0 ± 5.6	Visual analogue scale (0-10)	No	7.6 ± 0.9/8.0 ± 0.7
Dailey et al., 201318	Crossover RCT	TENS/placebo TENS/control	41/=/=	ACR 1990	97.6/=/=	49.1 ± 12.9/=/=	Visual analogue scale (0-10)	Yes	5.0 ± 0.5/5.0 ± 0.4/5.2 ± 0.4
Dailey et al., 202020	RCT	TENS/placebo TENS/control	103/99/99	ACR 1990	100/100/100	44.7 ± 14.3/47.2 ± 12.6/48.6 ± 11.8	Visual analogue scale (0-10)	Yes	6.2 ± 1.5/5.9 ± 1.4/6.1 ± 1.6
Jamison et al., 202136	RCT	TENS/placebo TENS	61/57	ACR 2010	93.5/93	52.3 ± 13.8/48.3 ± 13.1	Fibromyalgia Impact Questionnaire pain item (0-10)	No	6.8 ± 2.0/6.2 ± 1.8
Lauretti et al., 201346	RCT	Single TENS/double TENS/placebo TENS	13/13/13	ACR 1990	91.7/100/90	32 ± 8/30 ± 12/35 ± 8	Visual analogue scale (0-10)	No	8.5 ± 1.0/8.5 ± 2.0/8.0 ± 2.0
Löfgren and Norrbrink, 200954	Cross-over RCT	TENS/superficial warmth	32/=	ACR 1990	100/=	41.5 ± 8.3/=	Numerical rating scale (0-100)	No	80.0 (60.0-90.0)*/77.5 (62.5-85.5)*
Mutlu et al., 201364	RCT	Exercise + TENS/exercise	30/30	ACR 1990	100/100	45.63 ± 9.1/43.3 ± 10.8	Myalgic pain score (0-54)	No	37 (50-20)*/32 (50-20)*
Ozen et al., 201967	RCT	TENS + ultrasounds + hot pack/acupuncture	22/22	ACR 1990	100/100	45.2 ± 9.4/47.9 ± 7.6	Visual analogue scale (0-10)	No	8.4 ± 1.7/7.7 ± 1.1
Yüksel et al., 2019100	RCT	TENS/acupuncture	21/21	ACR NR	NR/NR	38.05 ± 11.3/44.6 ± 10.34	Visual analogue scale (0-10)	Yes	5.19 ± 2.20/4.62 ± 1.56

^*Interquartile range.

=, same patients in a crossover trial design; ACR, American College of Rheumatology; NR, not reported; RCT; randomized controlled trial.

Regarding the TENS parameters applied in each included trial (Table 2), high-stimulation frequencies (greater than 10 Hz to 200 Hz) were used in 7 studies of the 11 studies,^{8,18,54,64,67,80,100} 3 used mixed frequencies (changing between low and high frequencies within the same treatment session),^20,36,46 and only 1 trial applied low-frequency TENS stimulation (10 Hz or less).⁹ Both high and mixed frequencies were considered appropriate TENS dosing, whereas low-frequency stimulation was considered an inappropriate dose. The intensity of stimulation was adjusted to elicit a feeling described as between “strong but comfortable” and “the highest tolerable nonpainful level” in 7 of the trials of the 11 studies.^{8,18,20,36,54,64,67} In 1 trial,⁴⁶ a fixed intensity 60 mA stimulation was applied to all the participants, an intensity considered very high, which we recorded as a strong elicited feeling. In another trial,⁷⁷ the authors reported that the participants were taught to maintain the elicited feeling throughout the application by constantly increasing the TENS intensity. All the abovementioned intensities were considered appropriate. Two studies of the 11 studies^9,100 reported low intensities, described as producing tingling or mild tingling sensations, and were considered inappropriate stimulations. Subjects in 4 of the 11 studies were able to choose, based on their reported pain experience, the electrode placement of the TENS at the site of pain,^18,54,64,77 as they considered appropriate. In 6 of the 11 studies, TENS was applied at the same electrode placement site for all the participants, regardless of their reported pain: over certain dermatomes,⁹ standard tender points,^8,36,67 or different paravertebral levels (cervical, thoracic, or lumbar).^20,100 Because fibromyalgia pain can affect a variety of different areas among patients, it was considered inappropriate to not allow participants to choose their treatment area. In one trial, one group received TENS at the site the individuals said that they felt the worst pain (considered appropriate), whereas another group were unable to choose the electrode location (considered inappropriate).⁴⁶ Transcutaneous electrical nerve stimulation treatments were applied over 10 or more sessions in 8 of the 11 studies (considered appropriate)^{8,20,36,46,54,64,67,80}; however, in the other 3 trials, this parameter was considered inappropriate, with one trial administering 6 sessions⁹ and 2 trials only 1 single session.^18,67

Table 2 - The transcutaneous electrical nerve stimulation parameters applied in each included trial.

Author, year	Frequency	Intensity described	Electrode placement	Number of sessions	Main findings regarding pain
Castro-Sánchez et al., 20209	2 Hz ✗	Adjusted to create a tingling sensation in the patients without causing discomfort. ✗	Dermatome corresponding to certain muscles (trapezius, latissimus dorsi, gluteus, quadriceps, and tibialis anterior). ✗	6 sessions over 7 wk ✗	TENS group showed no significant differences. Dry needling group showed a significant reduction (P = 0.001)
Carbonario et al., 20138	150 Hz ✓	Strong but comfortable and without muscular contraction. ✓	Bilateral tender points of the trapezium and supraspinatus. ✗	16 sessions over 8 wk ✓	TENS group showed a significant reduction (P < 0.01). No significant differences in the no-TENS group
Da Silva et al., 200877	15 Hz ✓	Determined by the individual, reporting constant tingling throughout the application period. ✓	Patient-referred tender points. ✓	10 sessions over 3 wk ✓	TENS group showed a significant reduction (P = 0.004). No significant differences were found in the no-TENS group
Dailey et al., 201318	100 Hz ✓	Highest tolerable intensity. ✓	Patient's preference between cervical thoracic or lumbar–sacral junctions. ✓	1 session. ✗	TENS group showed no significant reduction, with no between-group differences between the placebo TENS or no TENS groups
Dailey et al., 202020	2-125 Hz ✓	Highest tolerable intensity. ✓	Cervicothoracic junction and lower back. ✗	28 sessions over 4 wk ✓	TENS group showed a significant reduction (P < 0.001) compared with the placebo group (P = 0.0006) and no TENS group (P < 0.0001)
Jamison et al., 202136	60-100 Hz ✓	Subjects instructed to maintain a strong but comfortable stimulation intensity. ✓	On the upper calf. ✗	Average of 68.9 ± 27.1 sessions over 3 mo ✓	TENS group showed a significant greater reduction (P < 0.05) compared with the placebo group
Lauretti et al., 201346	2-100 Hz ✓ 2–100 Hz ✓	60 mA ✓ 60 mA ✓	Lower back and centrally above and below the space between the C7 and T1 spinous processes. ✗ Patients referred the worst area of pain between low back or cervical area. ✓	14 sessions over 1 wk ✓ 14 sessions over 1 wk ✓	TENS group showed a significant reduction compared with the placebo group (P < 0.02) TENS group showed a significant reduction compared with the placebo group (P < 0.05)
Löfgren and Norrbrink, 200954	80 Hz ✓	Intensity equivalent to a strong but not unpleasant level. ✓	Patients referred the worst area of pain between low back or cervical area. ✓	21 sessions over 7 wk ✓	TENS showed a significant reduction (P < 0.05). No between-group differences with superficial warmth
Mutlu et al., 201364	80 Hz ✓	To the patient's tolerance level. ✓	Patient-referred painful sites. ✓	14 sessions over 7 wk ✓	TENS group showed a significant reduction (P < 0.01) compared with the no-TENS group
Ozen et al., 201967	100 Hz ✓	Adjusted to the patient's tolerability. ✓	Bilaterally: suboccipital muscle insertions, trapezius, above the scapula spine, and paraspinous 3 cm lateral to the midline at the level of the midscapula. ✗	15 sessions over 3 wk ✓	TENS showed a significant reduction (P < 0.002). No between-group differences to acupuncture
Yüksel et al., 2019100	70 Hz ✓	Finely adjusted to create a mild tingling feeling in the patients without causing contraction or discomfort. ✗	T2 and T6 paravertebral level in the upper thoracic region. ✗	1 session. ✗	TENS showed a significant reduction (P = 0.001). No between-group differences to acupuncture

✓, Appropriate dose; ✗, Inappropriate dose.

3.2. Methodological quality and risk of bias assessment

3.2.1. The PEDro scale

The methodological quality of the 11 studies included in this meta-analysis was assessed through the PEDro scale (Table 3). The item that presented the lowest quality was the blinding of the therapists because all but one of the studies (Ref. 36) did not meet these criteria. Most studies presented moderate to high methodological quality,^{9,18,20,36,54,64,67} except for 3 of them that presented acceptable or low methodological quality levels.^8,46,80

Table 3 - Assessment of the quality of the studies based on the PEDro scale.

Items
	1	2	3	4	5	6	7	8	9	10	11	Total
Castro-Sánchez et al., 20209		1	1	1	0	0	1	1	1	1	1	8
Carbonario et al., 20138		0	1	0	0	0	0	1	0	1	1	4
Da Silva et al., 200877		1	0	1	0	0	0	1	0	0	1	4
Dailey et al., 201318		1	1	1	1	0	1	1	1	1	1	9
Dailey et al., 202020		1	1	1	1	0	1	1	1	1	1	9
Jamison et al., 202136		1	1	1	1	1	1	1	1	1	1	10
Lauretti et al., 201346		1	0	1	0	0	0	1	0	1	1	5
Löfgren and Norrbrink, 200954		1	0	1	0	0	0	1	1	1	1	6
Mutlu et al., 201364		1	0	1	0	0	1	1	0	1	1	6
Ozen et al., 201967		1	1	1	0	0	1	1	1	1	1	8
Yüksel et al., 2019100		0	0	1	0	0	0	1	1	1	1	6

PEDro, Physiotherapy Evidence Database. 1: participant choice criteria are specified; 2: random assignment of participants to groups; 3: hidden assignment; 4: groups were similar at baseline; 5: all participants were blinded; 6: all therapists were blinded; 7: all evaluators were blinded; 8: measurement of at least one of the key outcomes was obtained from more than 85% of baseline participants; 9: intention-to-treat analysis was performed; 10: results from statistical comparisons between groups were reported for at least one key outcome; 11: the study provides point and variability measures for at least one key outcome.

3.2.2. RoB-2 assessment

The 11 RCTs included in this review were evaluated using the Risk of Bias 2 (RoB 2) tool. In the summary of the findings, the domain that presented the highest risk of bias was measurement of the outcome, with 36% of the studies showing a considerable risk of bias. Furthermore, 56% of the studies presented some concerns in relation to the randomization of the participants, and 72% of them presented certain risks because of deviations from the intended intervention. All studies were at a minimal risk of bias due to missing outcome data or selection of the reported results. Nevertheless, 54% of the studies were considered to have an elevated risk of bias in the overall assessment. The RoB-2 summary results and risk of bias graph are shown in Figures 2 and 3. The interrater reliability of the risk of bias assessment was high (κ = 0.785).

Figure 2.: Summary results of the Risk of Bias 2 (RoB 2) tool assessment.

Figure 3.: Risk of bias graph according to the Risk of Bias 2 (RoB 2) tool assessment.

3.3. Effects of transcutaneous electrical nerve stimulation on pain

Figure 4 presents the forest plot for the meta-analysis of the pooled standardized mean change difference in pain. Of note, considering all the studies included together, and not taking into account the TENS dosage applied, TENS had no mean effect on pain (d₊ = 0.51, P >0.050, k = 14). Nonetheless, there was large variability among the effect sizes (Q(13) = 167.50, P < 0.0001, I² = 94.9%).

Figure 4.: Forest plot of the meta-analysis for assessing the efficacy of TENS to alleviate pain. RE, random effects; TENS, transcutaneous electrical nerve stimulation.

3.4. Effects of transcutaneous electrical nerve stimulation dosage on pain

We used mixed-effects ANOVAs to analyze the potential different effects of the TENS dosage, considering the TENS parameters (frequency, intensity, and application site) and the number of treatment sessions. In these analyses, the dependent variable was the d index for pain reduction, and we also used 4 categorical moderator variables (Table 4). As shown in Table 4, there was a significant association with the effect sizes for 3 categorical moderator variables: the number of sessions (P = 0.005), frequency (P = 0.011), and TENS intensity (P = 0.047). The electrode placement did not show a statistically significant association with the effect sizes (P = 0.691).

Table 4 - Results of the weighted analysis of variances for the influence of categorical variables on the effect sizes.

Moderator variable	k	d +	95% CI		ANOVA results
			dl	du
No. of sessions					F(1,12) = 11.74, P = 0.005
Appropriate	10	0.942	0.374	1.510	R 2 = 0.553
Inappropriate	4	−0.692	−1.563	0.178	Q W(12) = 83.69, P < 0.001
Frequency					F(1,12) = 9.18, P = 0.011
Appropriate	13	0.661	0.135	1.187	R 2 = 0.542
Inappropriate	1	−1.937	−3.730	−0.144	Q W(12) = 67.07, P < 0.001
Intensity					F(1,12) = 4.93, P = 0.047
Appropriate	12	0.734	0.106	1.361	R 2 = 0.311
Inappropriate	2	−0.888	−2.351	0.575	Q W(12) = 100.95, P < 0.001
Electrode placement					F(1,12) = 0.17, P = 0.691
Appropriate	6	0.671	−0.430	1.772	R 2 = 0.0
Inappropriate	8	0.406	−0.492	1.304	Q W(12) = 166.91, P < 0.001

95% CI, 95% confidence interval; ANOVA, analysis of variance; d₊, pooled standardized mean change difference; dl, lower confidence limit for d₊; du, upper confidence limit for d₊; F, Knapp–Hartung statistic for testing the significance of the moderator variable; k, number of comparisons between treatment and control group; Q_W, statistic for testing the model misspecification; R², proportion of variance accounted for by the moderator.

The number of TENS sessions applied accounted for 55.3% of the variance (P = 0.005). Specifically, the studies that had completed an appropriate number of TENS sessions obtained a large mean pain reduction in individuals with FM (d₊ = 0.942, k = 10), whereas the studies that had used a suboptimal number of sessions did not achieve a pain relief effect in these individuals. Furthermore, the frequency accounted for 54.2% of the variance (P = 0.011). As shown in Table 4, studies that had used an appropriate TENS frequency obtained a moderate-to-large mean pain reduction (d₊ = 0.661, k = 13), whereas those that had used a suboptimal frequency observed an increase in the mean levels of pain (d₊ = 1.94, k = 1). Moreover, the TENS intensity was significantly associated with the effect size (P = 0.047) and accounted for 31.1% of the variance. In this case, the studies that used an appropriate intensity achieved a substantial reduction in pain (d₊ = 0.73, k = 12), whereas that those did not saw no change in the levels of pain.

Finally, we performed separate moderator analyses with the studies that had used an appropriate number of TENS sessions (k = 10), intensity (k = 12), or frequency (k = 13) to assess the potential difference these might have had on the effects of the TENS dosage. Given that in all the studies that had applied the appropriate number of sessions, the intensity and frequency variables had also been appropriate, the moderator analysis with studies that had used an appropriate number of TENS sessions (k = 10) was only carried out considering the TENS electrode placement. As shown in Table 5, only 1 moderator variable showed a significant relationship with the effect sizes: an appropriate number of sessions. This indicated that there were differences in the efficacy of TENS to reduce pain in the studies, which had or had not administered an appropriate number of sessions. Specifically, studies that had implemented an appropriate number of sessions achieved a higher pain relief effect compared with those that had not, with this effect accounting for more of 50% of the variance (P < 0.05).

Table 5 - Results of the weighted analysis of variances for the influence of categorical variables on the effect sizes in studies that had used an appropriate number of transcutaneous electrical nerve stimulation sessions, intensity, and frequency.

Moderator variable	k	d +	95% CI		ANOVA results
			dl	du
No. of sessions (k = 10)
Electrode placement					F(1,8) = 0.47, P = 0.512
Appropriate	4	1.156	0.179	2.133	R 2 = 0.0
Inappropriate	6	0.801	0.116	1.486	Q W(8) = 39.87, P < 0.001
Intensity (k = 12)
Electrode placement					F(1,10) = 0.17, P = 0.693
Appropriate	6	0.599	−0.308	1.505	R 2 = 0.0
Inappropriate	6	0.823	−0.006	1.652	Q W(10) = 63.04, P < 0.001
Number of sessions					F(1,10) = 6.15, P = 0.033
Appropriate	10	0.895	0.404	1.386	R 2 = 0.498
Inappropriate	2	−0.468	−1.590	0.654	Q W(10)= 44.43, P < 0.001
Frequency (k = 13)
Electrode placement					F(1,11) = 0.07, P = 0.801
Appropriate	6	0.590	−0.279	1.458	R 2 = 0.0
Inappropriate	7	0.723	−0.008	1.454	Q W(11) = 66.83, P < 0.001
Number of sessions					F(1,11) = 6.38, P = 0.028
Appropriate	10	0.891	0.418	1.364	R 2 = 0.471
Inappropriate	3	−0.230	−1.084	0.624	Q W(11) = 47.46, P < 0.001
Intensity					F(1,11) = 0.39, P = 0.547
Appropriate	12	0.714	0.139	1.290	R 2 = 0.0
Inappropriate	1	0.159	−1.721	2.039	Q W(11) = 63.60, P < 0.001

95% CI, 95% confidence interval; ANOVA, analysis of variance; d₊, pooled standardized mean change difference; dl, lower confidence limit for d₊; du, upper confidence limit for d₊; F, Knapp–Hartung statistic for testing the significance of the moderator variable; k, number of comparisons between treatment and control group; Q_W, statistic for testing the model misspecification; R², proportion of variance accounted for by the moderator.

3.5. Publication bias

The Egger test applied to the intercept of a simple regression model of the effect sizes did not reach statistical significance [t(12) = 0.068, P = 0.947]. Therefore, based on the results of these different analyses, publication bias can be reasonably discarded as a serious threat to our meta-analytic findings.

4. Discussion

This study showed that, from the perspective of a dose–response paradigm, the application parameters must be adequately selected for TENS to reach clinical effectiveness in FM. Thus, regardless of the electrode placement, the specific application of high frequencies (greater than 10 Hz to 200 Hz) or mixed frequencies (changing between low and high frequencies within the same treatment session), at a high intensity, for a minimum of 10 treatment sessions in individuals with FM resulted in a large mean reduction in pain compared with other parameter combinations.

However, so far, reviews of the published academic studies on TENS in patients with FM have not supported its use for individuals with FM.^26,41,74 Until this, only the meta-analysis study by da Silva et al. (2017), which included 6 studies on TENS in individuals with FM had been published.⁷⁴ The statistical analysis was performed by pooling the results from all 6 studies, without considering whether the applied dose had been appropriate or not, and did not show any significant differences in pain relief when compared with the control groups. In this work, we located 11 studies that allowed further exploration of the efficacy of TENS. When performing the same analysis previously completed in the meta-analyses on TENS, that is, pooling all the results from all the included studies, we obtained similar results to the previous meta-analyses indicating that TENS had no significant effect on pain. Hence, we decided to study the different effects of the TENS dosing parameters: frequency, intensity, application site, and number of sessions. Thus, to the best of our knowledge, this is the first review and meta-analysis of TENS on FM that, in addition to assessing the methodological quality criteria of the work, also considered what TENS application parameters were used.

According to our results, studies that applied an appropriate number of sessions (≥10) achieved a large mean pain reduction, whereas those that applied fewer sessions (<10) did not obtain the same pain relief. Thus, in our opinion, applying in FM at least 10 sessions should be considered appropriate dosing, whereas the administration of fewer sessions could be considered underdosing. Logically, receiving more sessions implies longer use of the treatment, which perhaps fits better to the chronic nature of the pain experienced by people with FM. However, in the literature, the repeated use of TENS at the same dose (regarding frequency or intensity, etc.) has been reported to lead to a decrease in the analgesic effects of TENS, both in animal and human studies.⁹³ This phenomenon is known as tolerance to TENS and is mediated through endogenous opioids, which decrease the analgesia produced by TENS treatment by raising the activation threshold in the mu and delta opioid receptors.^10,52 Nevertheless, this tolerance can be prevented by avoiding a repetitive pattern of electrical stimulation. This is achieved by using a mixture of frequencies within the same session to modify the electrical stimulus applied²² or by increasing the TENS intensity within sessions.⁷⁵ In all the trials, we included that had applied at least 10 TENS sessions, except for one, the intensity had been individually adjusted in each session and did not refer to a specific value (intensity in mA). Rather, they referred to a subjective perception (eg, “increase the intensity until you perceive a strong but comfortable sensation”). Thus, if from one session to another, the patient experienced a certain degree of tolerance to the TENS, manifested as a lower evoked perception at the same intensity, based on the instructions received, the patient would increase the intensity, thereby preventing the decrease in analgesic response due to tolerance. Finally, in the study applying the same intensity during all sessions, mixed frequency (2-100 Hz) was applied,⁴⁶ which is considered another means to prevent the appearance of tolerance.²² Thus, we inferred that the use of these 2 tolerance prevention mechanisms could have been effective in trials administering an appropriate number of sessions (≥10).

Regarding the frequency of the TENS applied, studies in which high-frequency or mixed frequencies were applied showed better pain relief results than the only one using a low frequency (10 Hz or less). Even though comparisons with a single study should be treated with great caution, there are several reasons that could explain this difference. First, high-frequency and low-frequency TENS activate different physiological pathways. High frequency acts through the delta opioid receptors, whereas low frequency involves the mu opioid receptor.^44,78,80 The literature contains evidence that suggests that long-term opioid treatment for chronic pain, which act through mu opioid receptors, may lead to opioid tolerance, which is also associated with less responsiveness to low-frequency TENS.^{10,44,51,78,79} Treatment with opioid medication in individuals with fibromyalgia continues to be widely used, as reported in recent studies by Jamison et al. 2022 (with 21.8% of FM individuals taking opioids), or in that of Dailey et al. 2022 (26%). High-frequency TENS, acting through delta opioid receptors, will likely be more appropriate in for individuals taking opioid analgesics due to cross-tolerance between low-frequency TENS and mu opioid receptors. Furthermore, clinically, patients typically find high-frequency TENS more comfortable than low-frequency TENS,⁸³ so that the therapy intensity can be increased more than in low-frequency TENS, thereby producing greater pain relief. Finally, as already mentioned, the use of mixed frequencies could help overcome the potential analgesic tolerance developed with long-term repeated application of TENS, allowing the pain relief effect to last longer.^10,19,52

In this study, the TENS intensity showed an association with the pain relief effect. There seems to be a consensus in the literature, both in experimental pain in healthy individuals and in clinical trials with individuals in pain, that the intensity of TENS must be high to achieve its greatest analgesic potential.^1,6,35,49,65 By contrast, several authors claim that TENS intensities that did not produce any feeling,^1,62,71 or that produced a low or mild sensation, would be ineffective.^1,13,15,62 Thus, patients must be instructed, so that they use the setting that produces the highest tolerable perceived intensity without pain, which should at least be perceived as “strong but comfortable.” This allows patients to adjust the intensity, even during a given session, to continue achieving the same level of intensity. This is because the sensation produced by TENS will fade during the treatment as a result of habituation of the nervous system to repetitive nonpainful electrical stimuli.³⁹

Regarding the electrode placement, our assumption about appropriateness was not correct as our findings indicated that was not significantly associated with the pain relief effect of TENS. Of note, the academic literature described possible sites for electrode placements other than those located at the site of pain or over nerve bundles. Indeed, electrodes have been positioned, with good results, at contralateral body sites when the cause of pain is a nervous system condition³⁹; paraspinal positions in individuals with visceral pain^2,53; or on acupuncture points remote to the site of pain.^39,93 The effectiveness of TENS applications in nonpainful areas could be explained by its widespread effect on the enhanced excitability of neurons in pain pathways^27,28,81 and in activating pain inhibitory mechanisms to reduce hyperalgesia^22,44,78 and allodynia.^12,65 Although the exact etiology of FM is currently unknown, the disease is associated with enhanced central excitability in the pain pathways^69,85 and loss of pain inhibition.^47,50,85 Therefore, if the other key parameters are applied accurately, the electrode position of TENS may be irrelevant in individuals with FM because the main analgesic mechanism of this treatment would not be generated through a local effect on the area of electrode application, but rather through a central system effect evoked regardless of this application area. In this sense, using TENS over the spine would be an appropriate placement because it would cover a large area including not only the spine but also the spinal nerves of the lower or upper extremities emerging from the spinal cord.

When separately analyzing results shown only in studies that had performed at least 10 TENS sessions, or that had used appropriate TENS intensities or frequencies to try to detect different effects of the TENS dosage, only one moderator variable showed a significant relationship to the effect sizes: the appropriate number of sessions. This would therefore suggest that the long-term application of TENS would be a prerequisite for its correct dosing in individuals with FM. This is consistent with previous studies on other chronic pain conditions in which TENS has been proven to produce a cumulative effect.^{6,11,25,48,60} Thus, even if the rest of the dose parameters (intensity, frequency, and electrode placement) were appropriate, not applying TENS therapy for a minimum length of time would prevent the clinical pain-reliving results potentially achievable by TENS therapy from being achieved.

When an appropriate dose of TENS treatment was applied, the positive findings related predominately to trials in which TENS had been administered in isolation (8 of the 11 studies) rather than as an adjunct to other physical therapy interventions (as in 3 of the 11 studies). This is clinically relevant because TENS could represent an analgesic resource separate from other physical therapeutic agents that patients could use at home to help them increase their activities of daily life.

In our study, we have not included the analysis of other parameters of the TENS electrical stimulus applied to the individuals. This is the case of the pulse duration because in the included studies, all the pulse durations used were very similar (150-290 µs), and no conclusions could have been drawn in this regard. Varying pulse durations has been shown to cause different physiological effects on motor unit recruitment in humans,³⁰ as well as in experimental studies in animals,^82,84 although there is no evidence that changes in pulse duration have significant effects on pain in humans.³⁹ Future studies will be necessary to ascertain if, in addition to the correct application of the parameters evidenced in this study, the variation of the impulse duration, or other pulse patterns, could increase the effectiveness of TENS.

A critical characteristic of TENS effectiveness studies is the timing of the outcome measure. In this regard, TENS provides the greatest effects while the unit is on or immediately after stimulation.^73,82,92 In our review, 8 of the 11 studies included did not measure pain when the peak effect was thought to occur. Moreover, TENS is thought to be more effective for movement-evoked pain rather than resting pain.^39,73 All the studies included in this meta-analysis measured resting pain, but only 2 of the 11 measured and compared both resting and movement pain. Dailey et al.¹⁸ in a study with a single session found that pain during movement was significantly lower during active TENS as compared with placebo or no TENS, whereas the resting pain showed no significant difference between the treatments. By contrast, Dailey et al.²⁰ found that after 4 weeks of active TENS, both movement-evoked pain and resting pain were significantly reduced compared with the placebo or no TENS group. The fact that most of the studies included did not measure pain when the peak TENS effects were supposed to occur (TENS on and measuring pain during movement) may have led to underestimating the TENS pain-relieving effect in these studies. Future TENS studies should thus be adequately designed, not only with the appropriate dosing but also with the proper timing of the outcome assessment, ie, with TENS on or immediately afterwards, during both resting and movement-evoked pain, to ensure not underestimating the TENS pain relief effect.

This study has certain limitations that should be pointed out. Our review and meta-analysis focused on the TENS effect on pain relief and did not study its effects on the patient function, quality of life, physical activity, or other associated symptoms, such as anxiety, fatigue, or sleep disturbance. It should be noted that about half (54%) of the included studies had a high risk of bias. This could influence the results found and be an explanation for the negative results found in some of the included clinical trials. Analyzing the RoB results in detail, it was found that the highest percentage of bias was found in the measurement of the outcome (domain 4). In this sense, we found that these studies did not correctly blind the participants, therapists, or evaluators. Despite the difficulty in blinding this type of intervention, it is necessary that future studies perform better masking protocols to ensure a lower risk of bias and confirm the results obtained currently. Besides, the limited number of RCTs included and the small sample size of most of them restricted the power of this study. Although a total of 789 individuals with FM from 11 studies were identified, and pain intensity data were pooled for 392 participants in the TENS arm, the threshold of acceptability of more than 500, suggested by Moore et al.,^59,60 to avoid the likelihood of producing imprecise effect estimations was not reached.

This review provides tentative suggestions regarding efficacious parameter selection for the use of TENS both in a clinically relevant pain management strategy for individuals with FM and as standardized doses to be used for testing in future clinical trials: a high frequency (greater than 10 Hz to 200 Hz) or mixed frequencies (changing between low and high frequencies within the same treatment session, ie, 2-100/125 Hz); a TENS intensity individually adjusted in each session to elicit a feeling between “strong but comfortable” to “the highest tolerable non-painful level”; and long-term intervention (at least 10 sessions). Future studies on TENS in individuals with FM should recruit large group arms, assess movement-evoked pain while TENS is on or immediately after treatment, and evaluate the long-term benefits of TENS on pain, related secondary variables, and its carryover effects. We also encourage the authors of future systematic reviews to use the principles outlined in this review to determine the efficacy of TENS and to exclude any studies not applying dose-related parameters associated with beneficial outcomes, even if they meet all the other methodological quality criteria in the study design.

Conflict of interest statement

The authors have no conflict of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at https://links.lww.com/PAIN/B786.