Search, screening and selection of results
The search of different databases identified 728 articles. In addition, 2 articles identified from the bibliographic citations of the selected articles. After the removal of duplicates, the titles and abstracts of 723 articles were analyzed to determine whether they met the inclusion criteria. After this second screening, which resulted in 696 articles being discarded because they dealt with subjects different from the focus of the study, 27 texts remained. Of these, 4 additional articles were excluded as reviews, and an additional 2 were excluded as case reports. Finally, 21 articles were included in the systematic review. The search, screening and selection process is reflected in the PRISMA flow chart (Fig. 1).
Description of included studies
In Additional file 1: Table S1 and Table 3, the characteristics of the 21 articles included in this systematic review are presented. Nineteen articles analyzed chronic effects of the WB-EMS, and 2 analyzed acute effects. Of the studies that analyzed chronic effects, 6 are part of a sequence of experimental phases that are called Test I [25], Test II [26]and Test III [2, 16, 17]. Test I [25] is described in a study. Test II [26] is documented by a study in German but is also detailed in a review [27] that the authors perform after these first two phases. Finally, three studies comprise Test III [2, 16, 17]. On five occasions, two or more articles refer to the same experimental phase: [28] with [29]; [17] with [16] and with [2]; [30] with [31]; [18] with [32]; [33] with [34, 35] with [36]. The remainder of the studies refer to independent experimental phases.
Characteristics of the sample
In the studies that are part of this review, a total of 505 subjects were analyzed, including 310 men and 195 women. A total of 178 were subjects with a certain level of training, whereas 327 were sedentary. For the most part, the studies that analyzed chronic effects are comprise samples from postmenopausal women. In Test I [25] (n = 30), the participants were trained women. In Test III [2, 16, 17] (n = 60) and [33,34,35,36] (n = 100), the participants were sedentary individuals with sarcopenia or/and osteopenia. In other studies [30, 31] (n = 75), the sample population suffered from sarcopenic obesity and metabolic syndrome. In Test II [26] (n = 28), the sample population included sedentary men with metabolic syndrome. In another study [37] (n = 41), the sample population included sedentary men but with a good health. In six other studies [38] (n = 9), [39] (n = 18), [40] (n = 26), [41] (n = 19) and [18, 32] (n = 20), the subjects were trained men. In [42] (n = 30 woman and n = 34 men), the participants where sedentary young people (20–25 years). Finally, in the experimental phase of Filipovic [28, 29] (n = 15), participants were professional soccer players. Due to the existence of articles from the same experimental phase, in this review, the subjects of each of these studies have been counted only once to avoid incurring a risk of bias.
Interventions
In the Test I [25], all the participants underwent two supervised sessions of 60 min weekly and another two sessions of 25 min at home (these sessions consisted of aerobic exercises, multilateral jumps and strength exercises 1–3 sets, 6–12 repetitions, 70–85% 1RM). In addition, the electrical stimulation group underwent a weekly training session with 15 exercises to strengthen the larger muscle groups. In Test II [26], the experimental group performed 15 min of elliptical exercise at 70–85% of the maximum aerobic speed in addition to 15 min strength exercises for the main muscle groups with a short range of movement. All of these exercises were performed with superimposed WB-EMS. The control group (CG) stretched on vibratory platforms in 18-min sessions with a frequency of 30 Hz, an amplitude of 1.7 mm and an acceleration of 1.3 to 2.2 g. In Test III [2, 16, 17], both the experimental and the control group underwent 10–14 dynamic exercises without additional load in each session (1–2 sets of 8 repetitions). The experimental group trained uninterruptedly during the study in three sessions every two weeks, whereas the CG trained a 60-min session weekly for 2 periods of 10 weeks separated by a 10-week period of inactivity. In the study by Kemmler et al. [37], high-intensity training (HIT) was compared with another training regimen with WB-EMS. The HIT consisted of sessions of 10/13 exercises between strength machines and core exercises. In the first two weeks, 2 sets of 15 repetitions were performed. In the following two weeks, two sets of 8–10 repetitions and in the remaining four weeks, muscle failure was addressed by further decreasing the number of repetitions per set from 8 to 3. The experimental group underwent 1–2 sets of 6–8 repetitions of 12 core-strengthening exercises with WB-EMS superimposed in standing position without an additional load.
Another experimental phase [30, 31] included a CG that did not undergo any type of training. The experimental group performed slight movements of the upper and lower limbs while in a half-lying supine position without additional load but with a superimposed WB-EMS. In this study, a second experimental group was included for which supplementation was provided. In Filipovic et al. [28, 29], the entire sample performed 3 sets of 10 repetitions of squats, but the experimental group did so with superimposed WB-EMS. In Wirtz et al. [18, 32], the entire sample performed 4 sets of 10 repetitions: the first at 50% of 10RM and the other three at 100% of 10RM. The only difference in their treatment was the application of the WB-EMS superimposed on the experimental group. In the study by Wolfgang Kemmler et al. [41] about acute effects on caloric expenditure, all the study subjects performed the same protocol that was already applied in Test I [25] with the exception that the experimental group performed it with WB-EMS superimposed without any additional burden. In Kemmler et al. [33,34,35,36], the experimental group performed the same exercises described in Test II with superimposed WB-EMS in addition to receiving a protein supplement. A second experimental group only received the protein supplementation. The control group did not perform any type of exercise and did not receive protein supplement. In Jee [42], the experimental group performed ten types of isometric exercises with WB-EMS superimposed, whereas the CG performed the same exercises without WB-EMS. In De la Cámara et al. [38], all participants performed the same training in three separate days under identical conditions, but different recovery methodologies were applied each day. One of these methodologies was the application of WB-EMS in the prone supine position.
Current parameters and intensity
In most studies, the frequency of the current was 85 Hz. In Test I [25], after 10 min with this frequency, 7 Hz was applied for an additional 10 min. In Test II [26], the applied frequency was the inverse as it involved 15 min at 7 Hz followed by 15 min at 85 Hz. In Kemmler et al. [33,34,35,36], the applied current was 85 Hz during the entire session. Amaro [39] applied the current following an undulating periodization model in which the frequency varied from 12 to 90 Hz. The chronaxie or pulse width was of 350 μs in all cases. The parameter that varied the most during the studies was the duty cycle, which indicates the relationship between contraction time and resting time. Although most studies involve 4–6 s of work every 4 s of rest, Filipovic et al. [28, 29] proposed 4 s of work every 10 s of rest. In contrast, Wirtz et al., [18, 32] proposed 5 s of work every 1 s of rest. The rise ramp is the time that elapses from the beginning of the electrical stimulus to its maximum intensity, where i was 0 s n all cases. The same occurred with the descent ramp. To understand the internal load that caused the current in the subjects, most of the studies used the Borg scale with the exception of Test I [25] and Test II [26]. In these tests, a scale 1 to 7 was used with 1 representing the lowest current intensity perception and 7 the highest. Wirtz et al. [18, 32] performed a test to understand the pain threshold to apply an intensity corresponding to 70% of said threshold of pain during the intervention. However, Wolfgang Kemmler et al. [37] applied the current to an intensity equivalent to “hard = 15” or “very hard = 17” on the Borg scale in which the maximum level is 20. In Jee [42], as the exercises were isometric, they were able to apply a current intensity corresponding to the maximum tolerance. All studies used the same electrical stimulator device (MIHA bodytec® (Augsburg, Germany) except for Jee [42], which used Miracle® suit (Seoul, Korea). Both devices generate a type of bipolar, rectangular and biphasic current.
Risk of bias
Figure 2 analyzes the different items used in the analysis of the risk of bias in each study. In Fig. 3, each type of risk of bias is studied at a general level.
Random sequence generation (selection bias)
All of the studies perform a randomization of the sample; however, in some cases, the methodology could incur some methodological error. In the case of Filipovic et al. [28, 29], there is a possible risk of selection. The author mentions that despite performing sample randomization, it allows a subject of the study to choose their membership in the CG given the discomfort that the WB-EMS imposes on them. In De la Cámara et al. [38] and Kemmler et al. [41], experimental and control groups include the same subjects who perform the intervention under different conditions, so the study is not truly randomized.
Allocation concealment (selection bias)
In Wolfgang Kemmler, et al. [40], the sample was decompensated due to the enormous numerical difference between the subjects that comprised the CG and the group of WB-EMS.
Blinding of participants and personnel (performance bias)
In the study by Filipovic et al. [28, 29], it is understood that if the subjects could choose their membership in the CG, it would be very likely that the entire sample knew the protocol of the study and the group to which they belonged. In such a case, there would not be a blinding of the participants with a possible placebo effect.
Blinding of outcome assessment (patient-reported outcomes)
In the studies by Filipovic et al. [28, 29] and Amaro et al. [39], there is no evidence that there was a blinding of the evaluators, so a risk of bias exists. On the other hand, in three additional studies [38, 40, 41], the crossover design was used, so the complete sample was at the same time. The sample of the control group, after a wash-out period, was the same in experimental group, so blinding of the evaluators was not possible.
Selective reporting (reporting bias)
The results are presented partially in different articles in six of the experimental phases that are analyzed in this systematic review: Filipovic et al. [28, 29], Test III [2, 16, 17], (W Kemmler et al. [30, 31], Wirtz et al. [18, 32], Kemmler [33,34,35,36] and Amaro et al. [39]. This method could incur a possible risk of notification bias given the possibility that it is mistakenly understood that these are different experimental phases, which would magnify the results of the same study.
In Jee [42], intragroup analysis is performed for psychophysiological variables but not for cardiopulmonary variables, which prevents the analyses of the effectiveness of WB-EMS in such variables.
Comparability of treatment and control group at entry
This type of risk of bias is more conflicted with the rigorous scientific procedure. In Test I [25], WB-EMS is applied in an extra weekly session to the experimental group in which the participants performed a series of exercises that were not practiced by the subjects of the CG, so it is impossible to objectively determine the isolated effect of WB-EMS. In the Test II [26], the WB-EMS is not the only differentiating variable in both groups because the CG performs stretching work on a vibration platform instead of performing the same exercises as the experimental group. Thus, it is not possible to determine the isolated effect of WB-EMS. The same limitations are noted in Kemmler [30]. In this study, the CG did not perform any exercise, whereas the experimental group performed upper and lower limb movements while electrical stimulation occurred. In Test III [2, 16, 17], it seems that the training of both groups is based on the same exercises. However, the WB-EMS group performed three sessions every two weeks, whereas the CG group completed a weekly session of 60 min per week and rested 10 weeks during the course of the study. Thus, the treatment is not equal in terms of volume and distribution of the loads in both groups. In the study by Wolfgang Kemmler [40], the group treatments were not the same because the objective was to compare the effect of two different activities. Thus, one group ran a marathon, and the other group underwent WB-EMS training. A similar experimental setup was noted in Amaro et al. [39]. In this study, the two experimental groups performed strength exercises during their weekly session of WB-EMS, but the control group exclusively performed aerobic running throughout the study. In the study by Kemmler [37], a group performed exercises in the context of WB-EMS that differed from those used by the CG that relied on high intensity training (HIT) with guided motion strength machines. This difference could perhaps allow comparisons of the effects of the two trainings but could not determine the adaptations that WB-EMS causes alone. Finally, in the study by Kemmler [33,34,35,36], the CG did not perform the same exercises as the experimental group (in fact, CG did not perform any type of exercise). Thus, it is not possible to determine whether the possible improvements are attributed to the exercises or to WB-EMS; thus, the effect of WB-EMS alone cannot be analyzed.
Outcome measures
Anthropometric parameters
Research on WB-EMS has identified minimal effects in relation to anthropometric parameters. In Filipovic et al. [29], Kemmler et al. [37], and Wirtz et al. [32], no significant changes were found. In Test I [25], body weight decreases (− 1.9 ± 1.7 kg, p = 0.001). However, body weight is also decreased in the CG (− 0.9 ± 1.5 kg, p = 0.025), and no significant differences are noted between both groups. No changes were noted in Test II [26] and Test III [2, 16, 17]. Total body fat is reduced in Kemmler et al. [33,34,35,36] (− 2.05 kg (− 1.40 to − 2.68), p = 0.001), but the difference between WB-EMS&P and the protein groups was borderline nonsignificant (p = 0.051). Regarding the sum of skinfolds, in Test I [25], a decrease is observed (− 8.6%, p = 0.001). However, the value increases (1.4%) albeit nonsignificantly in the CG. The waist circumference is reduced in this same study (− 2.3%, p = 0.001), whereas an increase is noted in the CG (1%, p = 0.106). The hip circumference decreases in Test I (− 2.3%, p = 0.001) and in the CG (1.3%, p = 0.008). The waist circumference decreases (− 5.7 ± 1.8 cm, p = 0.001) in Test II [26] and in the CG (− 3.0 ± 2.0 cm, p = 0.006). In Test III [2, 16, 17], the waist circumference decreases (− 1.1 ± 2.1 cm). However, a large deviation is observed, which is also noted in the increase observed in the CG (1.0 ± 2.8 cm). The level of significance of these data is not provided. Similar findings are noted in Kemmler et al. [33,34,35,36] where waist circumference decreases (− 1.94 cm (− 1.44 to − 2.44), p = 0.001) with a significant group difference (p = 0.001) between the treatment group and the CG (− 0.10 cm (.46 to −.67)). A high deviation also occurs in the study by Kemmler et al. [31], where waist circumference is reduced in the WB-EMS group (− 1.5 ± 2.3%, p = 0.004) and the CG (− 0.02 ± 2.26%, p = 0.963). Muscle mass increases in Test I [25], Test II [26], Test III [2, 16, 17], Kemmler et al. [31] and Kemmler et al. [33,34,35,36]. However, in all cases, the effect is minimal with large deviations and a low level of significance. Similar results were noted for appendicular muscle mass in Test I [25], Test II [26] and Test III [2, 16, 17]. Regarding fat mass, Additional file 2: Table S2 demonstrates that the changes are almost imperceptible, and large deviations are noted in Test II [26], Test III [2, 16, 17] and Kemmler et al. [31]. In Test III [2, 16, 17], the evolution of bone mineral mass is measured but no effects were observed.
Strength parameters
Filipovic et al. [28, 29] was the only study that measured the 1RM, observing an increase of 22.42 ± 12.79% (p < 0.01) after fourteen weeks of WB-EMS without changes in the CG. According to the authors, this gain in strength explains the improvement in sports skills, such as the linear 5-m sprint (− 0.3 s, p = 0.039), 10-m sprint with changes of direction (− 0.18 s, p = 0.024), one-step chute speed (+ 9.9 km/h, p = 0.001), and squat jump (+ 2.9 cm, p = 0.021). Most of the measurements that are made to study the evolution of strength analyze its manifestation in the isometric muscle contraction regime. In Test I [25], the isometric maximal strength improved (9.9%; p = 0.015) in the extensors of the leg and extensors of the trunk (9.6%; p = 0.001), which is parameter that was reduced in the CG (− 6.4%, p = 0.054 and-4.5%, p = 0.106). In Test II [26], improvements in power (+ 10 ± 7%, p = 0.01) and isometric maximal strength (+ 15 ± 11%; p = 0.01) of the leg extensors were observed, whereas both parameters decreased (+ 3 ± 4% and − 0.5 ± 6%) in a nonsignificant manner in the CG (p-values not provided). Increases were noted in Test III [2, 16, 17] (9.1 ± 11.2%, p = 0.002) and the CG (1.0 ± 8.1%, p = 0.631). However, large standard deviation was noted and the data lacked significance. Similar results were noted in Kemmler et al. [37] as presented in Additional file 2 Table S5. Handgrip strength increased in the study by Kemmler et al. [33,34,35,36] (1.9 kg (0.99 to 2.82), p = 0.001)) with a small size effect and large deviations, and non-significant differences were observed between the treatment group and the CG. In the same study, maximum dynamic strength “leg-press” increases (189 ± 129 N, p = 0.001), but the difference between WB-EMS&P and the protein group was not significant.
Energy expenditure and cardiovascular system
Kemmler [41] conducted a study of caloric expenditure by indirect calorimetry of a 16-min session of low intensity strength exercises performed by young subjects (26.4 ± 4.3 years), revealing an increase of 17% with superimposed WB-EMS (412 ± 60 kcal · h-1 versus 352 ± 70 kcal · h-1, p < 0.01). However, in Test I [25], no significant increase in resting metabolic rate was observed after 14 weeks of training.
Blood parameters
Filipovic et al. [28, 29] did not report significant differences in the evolution of blood parameters, such as the concentration of red blood cells, platelets, white blood cells or hemoglobin. The authors of this experimental phase report that in week 7 and 14 of their intervention, an increase (p < 0.05) in the size and deformability of red blood cells was observed. These results indicate an increased capacity for the transport of oxygen to muscular cells. However, the effect size is not recorded. In Kemmler et al. [31], no changes in triglycerides, glucose and cholesterol were observed after 26 weeks of training with WB-EMS. Similar results were noted in Wirtz et al. [18] given that no differences were noted in the analysis of the evolution of testosterone, cortisol and growth hormone. A positive aspect of this study is the absence of pre-post changes in other parameters that could indicate overtraining in cases with observed high values, such as lactic acid and creatine kinase (CK) activity. Kemmler et al. [33,34,35,36] observed significant changes in the total cholesterol/HDL-C ratio (− 0.31 index (−.15 to −.47), p = 0.001) and in the protein group who did not train with superimposed WB-EMS.
Psychophysiological parameters
Jee [42] is observed a positive effect of WB-EMS in psychophysiological variables using a scale from 1 to 10. Soreness (− 4.16 ± 1.20), anxiety (− 3.75 ± 0.91), fatigability (− 3.33 ± 1.01) and sleeplessness (− 4.88 ± 1.13) were significantly changed (p = 0.001). Control group data were not provided.