Electromyographic analysis of the serratus anterior and upper trapezius in closed kinetic chain exercises performed on different unstable support surfaces: a systematic review and meta-analysis

Introduction

Scapular muscle imbalance is a frequently reported alteration in individuals with shoulder pain, overhead athletes, and physically active and healthy populations (Ludewig & Cook, 2000; Cools et al., 2007; Cools et al., 2014; De Mey et al., 2013; Kinsella & Pizzari, 2017). A correct muscle balance between the scapular muscles (e.g., upper trapezius (UT), lower trapezius (LT), and serratus anterior (SA)) is important for normal shoulder function (Ludewig & Cook, 2000; Cools et al., 2007). For instance, a decrease in the activation of the SA or LT, and the consequent overactivation of the UT, generates an alteration of the position and scapular movement, i.e., dyskinesia, characterized by excessive upward rotation and anterior scapular tilt during arm elevation which can lead to subacromial impingement or glenohumeral instability (Ludewig & Cook, 2000; Kibler et al., 2013; Kinsella & Pizzari, 2017; Huang, Ou & Lin, 2019).

It is widely recognized that scapular dyskinesis rehabilitation programs focused on decreasing UT activation and increasing SA and LT activation in early stages of the rehabilitation process contribute to restoring normal scapular position and movement (Cools et al., 2007; Youdas et al., 2020b; Berckmans et al., 2021). Several closed kinetic chain exercises have been reported as efficient to increase SA and LT myoelectrical activity, such as “push-up” (Calatayud et al., 2014a; Santos et al., 2018), “push-up plus” (scapular protraction posture in the end of the ascending phase of a push-up) (Cools et al., 2014; Batbayar et al., 2015; Gioftsos et al., 2016; Hwang et al., 2017), “scapular protraction” (scapular protraction and retraction with elbows extended) (Andersen et al., 2012; Lee, Lee & Park, 2013; De Mey et al., 2014), and “plank” (prone position, with the trunk and leg fully extended, the shoulders flexed at 90° and elbows in extension) (Pontillo et al., 2007; Tucker et al., 2010; Ashnagar et al., 2016; Oliver et al., 2018).

As a mean of progression in UT and SA strengthening and rehabilitation programs, unstable support surfaces have been incorporated into the push-up variants. Compared to stable surface, increased SA and UT EMG during push-ups has been reported when performed on unstable surfaces such as both-sides-up (BOSU) ball (Tucker et al., 2010; Borreani et al., 2015a), wobble board (Park & Yoo, 2011; Biscarini, Contemori & Grolla, 2019), therapeutic ball (Seo et al., 2013), and suspension equipment (Jeong, Chung & Shim, 2014; De Mey et al., 2014). However, contradictory results have been reported, observing a significant decrease or no difference in the EMG of the SA when comparing different types of unstable surfaces against stable surface (Pirauá et al., 2014; Gioftsos et al., 2016; Horsak et al., 2017).

In sport rehabilitation, devices with a greater base of support (e.g., BOSU ball) are used as a progression to stable surfaces in earlier stages of rehabilitation or training, while those with a smaller base (e.g., suspension equipment) or greater mobility (e.g., therapeutic ball) are integrated in more advanced stages (Behm et al., 2010). In this context, there is evidence that suspension mechanisms and the therapeutic ball could generate a greater neuromuscular recruitment demand compared to other support surfaces such as BOSU or floor (Lehman, Hoda & Oliver, 2005; Imai et al., 2010; Behm & Colado, 2012; Czaprowski et al., 2014; Borreani et al., 2015a; Youdas et al., 2020b). However, the comparative analyzes observed in previous reviews (Kang et al., 2019; Cappato de Araújo et al., 2021) have grouped in a single data pool different types of unstable support surfaces—BOSU, therapeutic ball, suspension equipment, among others—without considering the possible and potential differences in neuromuscular demand induced by the individual analysis of each unstable support surface (Mendez-Rebolledo et al., 2021). In this context, to our knowledge, there is no quantitative analysis of the evidence that groups the data according to these differences. In summary, previous reviews chose to analyze EMG activity in subgroups that considered the type of exercise (push-up, push-up plus, knee push-up, among others), type of execution (isometric and dynamic), and type of surface (stable and unstable). However, the influence each type of unstable surface (BOSU, wobble board, therapeutic ball, or suspension equipment) on the EMG activity still unclear. Therefore, the purpose of this systematic review with meta-analysis was to analyze the individual effect of different unstable support surfaces (i.e., BOSU ball, wobble-board, therapeutic ball, and suspension equipment) compared to a stable support surface on scapular muscles EMG activity during the execution of closed kinetic chain exercises.

Survey methodology

Study design

This systematic review and meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) updated to year 2020 (Page et al., 2021). The whole process of study selections was summarized in the PRISMA flow diagram (Fig. 1).

Results

Quantitative synthesis: meta-analysis

For the meta-analysis, the articles were grouped considering the types of support surface [stable (floor) compared to wobble board, BOSU ball, therapeutic ball, or suspension equipment] for each scapular muscle (SA and UT) and push-up variants, since at least three studies provided sufficient data for the calculation of ES. Finally, twenty-three low-to-high quality studies (498 participants) were included in the meta-analysis.

Upper trapezius

Stable surface compared to suspension equipment.

The comparison involved six studies (Lee, Lee & Park, 2013; Calatayud et al., 2014b; Calatayud et al., 2014a; De Mey et al., 2014; Borreani et al., 2015b; Horsak et al., 2017), involving 6 groups that completed push-ups on stable surface (n = 163), and 6 groups that completed push-ups on suspension equipment (n = 163). The forest plot (Fig. 2A) revealed greater UT EMG activity during suspension equipment compared to stable surface (ES = 2.92, very large; 95% CI [0.92–4.92]; p = 0.004; I² = 97.6%; Egger’s test p = 0.055). After a sensibility analysis with each study removed once from the meta-analysis to assess its impact on results heterogeneity, the results remained similar, with I² values > 90%. Moreover, a sensitivity analysis was conducted using the trim and fill method, and the results remained similar, with an ES = 2.06 and 95% CI [0.11–4.01].

Stable surface compared to wobble board.

The comparison involved eight studies (Maenhout et al., 2010; Park et al., 2013; Park & Yoo, 2013; Pirauá et al., 2014; Gioftsos et al., 2016; De Araújo et al., 2018; De Faria et al., 2021; Patselas et al., 2021), involving 8 groups that completed push-ups on stable surface (n = 150), and 8 groups that completed push-ups on wobble board (n = 150). The forest plot (Fig. 2B) showed greater UT EMG activity during wobble board compared to stable surface (ES = 0.33, small; 95% CI [0.11–0.56]; p = 0.003; I² = 0.0%; Egger’s test p = 0.347).

Stable surface compared to BOSU ball.

The comparison involved three studies (Tucker et al., 2010; Karagiannakis, Athanasopoulos & Mandalidis, 2018; De Araújo et al., 2020), involving 3 groups that completed push-ups on a stable surface (n = 53) and 3 groups that completed push-ups on BOSU ball (n = 53). The forest plot (Fig. 2C) showed similar UT EMG activity during BOSU ball compared to stable surface (ES = 0.30, small; 95% CI [−0.08–0.67]; p = 0.119; I² = 0.0%; Egger’s test p = 0.123).

Stable surface compared to therapeutic ball.

The comparison involved three studies (Lehman, Gilas & Patel, 2008; Sandhu, Mahajan & Shenoy, 2008; Seo et al., 2013), involving 3 groups that completed push-ups on stable surface (n = 55) and 3 groups that completed push-ups on therapeutic ball (n = 55). The forest plot (Fig. 2D) showed greater UT EMG activity during therapeutic ball compared to stable surface (ES = 1.03, moderate; 95% CI [0.64–1.42]; p < 0.001; I² = 0.0%; Egger’s test p = 0.753).

Serratus anterior

Stable surface compared to suspension equipment.

The comparison involved six studies (Lee, Lee & Park, 2013; Calatayud et al., 2014b; De Mey et al., 2014; Borreani et al., 2015b; Horsak et al., 2017; Youdas et al., 2020a), involving six groups that completed push-ups on stable surface (n = 167) and 6 groups that completed push-ups on suspension equipment (n = 168). The forest plot (Fig. 3A) showed similar SA EMG activity during suspension equipment compared to stable surface (ES = −0.03, trivial; 95% CI [−1.74–1.69]; p = 0.978; I² = 97.5%; Egger’s test p = 0.859). After a sensibility analysis with each study removed once from the meta-analysis to assess its impact on results heterogeneity, the results remained similar, with I² values > 90%. Nonetheless, when a sensitivity analysis was conducted using the trim and fill method, the results indicated an ES = 1.54 and 95% CI [−0.54–3.62], favoring greater EMG for suspension equipment compared to stable surface.

Stable surface compared to wobble board.

The comparison involved nine studies (Maenhout et al., 2010; Park et al., 2013; Park & Yoo, 2013; Pirauá et al., 2014; Borreani et al., 2015a; Gioftsos et al., 2016; De Araújo et al., 2018; De Faria et al., 2021; Patselas et al., 2021), involving nine groups that completed push-ups on stable surface (n = 180) and 9 groups that completed push-ups on wobble board (n = 180). The forest plot (Fig. 3B) showed similar SA EMG activity during wobble board compared to stable surface (ES = 0.57, small; 95% CI [−0.26–1.40]; p = 0.177; I² = 92.6%; Egger’s test p = 0.007). After a sensibility analysis with each study removed once from the meta-analysis to assess its impact on results heterogeneity, the results remained similar, with I² values > 90%, except when a study was removed (Borreani et al., 2015a), as the I² reached a value = 0.0%. Nonetheless, the p value remained non-significant (p = 0.276). The ES changed to −0.12 and the Egger’s test to p = 0.053, with values adjusted to ES = −0.25 and 95% CI [−0.46–−0.04].

Stable surface compared BOSU ball.

The comparison involved five studies (Tucker et al., 2010; Kim et al., 2014; Borreani et al., 2015a; Karagiannakis, Athanasopoulos & Mandalidis, 2018; De Araújo et al., 2020), involving five groups that completed push-ups on a stable surface (n = 98) and 5 groups that completed push-ups on a BOSU ball (n = 98). The forest plot (Fig. 3C) showed similar SA EMG activity during inverted Bosu compared to stable surface (ES = 1.47, large; 95% CI [−0.31–3.24]; p = 0.106; I² = 96.4%; Egger’s test p = 0.015). After a sensibility analysis with each study removed once from the meta-analysis to assess its impact on results heterogeneity, the results remained similar, with I² values > 90%, except when a study was removed (Borreani et al., 2015a), as the I² reached a value = 74.4% (moderate). Nonetheless, the p value remained non-significant (p = 0.824). The ES changed to 0.08 and the Egger’s test p = 0.405, with values adjusted to ES = 0.21 and 95% CI [−0.39–0.81].

Stable surface compared to therapeutic ball.

The comparison involved three studies (Lehman, Gilas & Patel, 2008; Sandhu, Mahajan & Shenoy, 2008; Seo et al., 2013), involving 3 groups that completed push-ups on stable surface (n = 50) and three groups that completed push-ups on therapeutic ball (n = 50). The forest plot (Fig. 3D) showed similar SA EMG activity during therapeutic ball compared to stable surface (ES = 0.06, trivial; 95% CI [−0.51–0.62]; p = 0.844; I² = 45.7%; Egger’s test p = 0.072).

Discussion

The systematic review of the evidence allowed to find 30 articles, in which EMG activity of SA and UT was analyzed during the execution of closed kinetic chain exercises, comparing a stable support surface with different types of unstable support surfaces. Unlike other studies, this meta-analysis is the first to compare different types of unstable support surfaces (e.g., BOSU ball, therapeutic ball, and wobble board) with a stable surface (e.g., floor), without combining them in a single pool of data.

The push-up variants (push-up, half push-up, and knee push-up) were the only exercises that fulfilled the criteria for the meta-analysis, showing in increasing order, a higher UT EMG activity when performed with wobble board (ES = 0.33), therapeutic ball (ES = 1.03) and suspension equipment (ES = 2.92) compared to a stable support surface, without observing any effect for the Bosu ball. According to GRADE recommendations (Guyatt et al., 2011; Zhang et al., 2019a; Zhang et al., 2019b), results must be interpreted with caution due to low quality of evidence of included studies, downgraded mostly by inconsistency (i.e., I² values > 75%) and imprecision (wide confidence intervals). Nonetheless, after sensitivity analyses, the results remained relatively consistent for all comparisons on UT myoelectrical activity. Based on the ES, these findings allow to establish a progression in the level of neuromuscular demand generated by different types of unstable support surfaces. This could be applied in the progressive prescription of scapular muscle training exercises and potentially in the rehabilitation of individuals with musculoskeletal dysfunctions of the shoulder complex.

From the point of view of the magnitude of the EMG activity, it was observed that the suspension equipment had an averaged increase of 9% of MVIC of the UT. This increase is higher than the observed in previous systematic reviews: Kang et al. (2019) showed a 2.85% MVIC increase and Cappato de Araújo et al. (2021) showed a 5.81% MVIC and 6.01% MVIC increase during push-up and half push-up, respectively. These differences could be explained because the articles included in the current meta-analyses were grouped according to the intrinsic characteristics of each instability device and its possible effects on neuromuscular recruitment.

In this way, the suspension equipment generates an asymmetric and higher shoulder complex instability, and additionally, a greater mobility of the proximal joints compared to distal joints, causing a greater displacement of the center of mass (Horsak et al., 2017), which seems to generate higher levels of EMG activation of the scapular and trunk stabilizer muscles compared with other types of unstable devices such as wobble board and therapeutic ball (Beach, Howarth & Callaghan, 2008; Anderson et al., 2013; Maeo et al., 2014; Calatayud et al., 2014a; Borreani et al., 2015a). The latter, on the contrary, have a bilateral symmetrical support, which would cause lesser demand on the scapular stabilizer muscles (Lehman et al., 2006). In addition, during the execution of a push-up with suspension equipment, the arm reaches around 90° of glenohumeral flexion accompanied by scapular elevation and upward rotation (De Mey et al., 2014; Borreani et al., 2015a). These scapular movements are caused by the action of UT (Kibler, Sciascia & Wilkes, 2012), which would justify the higher myoelectric activity observed in this muscle. On the other hand, UT had a higher EMG activity level on a therapeutic ball compared to a stable surface. In this sense, the therapeutic ball could generate a stimulus that would request a greater pushing force towards the ground to the distal muscles (e.g., triceps brachii and biceps brachii) over the proximal muscles (e.g., UT and SA) of the upper limb, due to the high deformation and mobility of the therapeutic ball under each hand (Lehman et al., 2006; Maeo et al., 2014; Bezerra et al., 2020), which could explain lower levels of neuromuscular activity compared to the suspension device.

On the other hand, no differences were identified in the SA EMG activity between different types of support surfaces. According to GRADE recommendations (Guyatt et al., 2011; Zhang et al., 2019a; Zhang et al., 2019b), results must be interpreted with caution due to imprecision, risk of bias, and inconsistency. In fact, after sensibility analyses using the trim and fill method, the results indicated greater SA EMG activity using suspension equipment compared to stable surface. In addition, after a study-by-study sensibility analyses, the removal of one study from the meta-analysis (Borreani et al., 2015a) indicated a small tendency toward greater EMG in stable surface compared to wobble board, although non-significant. This can be explained by an extremely low mean difference with respect to the rest of the studies, however, in the risk of bias analysis, no differences were observed in the internal validity items that would justify the exclusion of the study from the analysis.

These findings are similar to those observed by Kang et al. (2019), who did not observe effect of unstable surfaces on SA neuromuscular activity. Conversely, in his recent review Cappato de Araújo et al. (2021) showed a decrease in EMG activity of the SA while including unstable surfaces in one arm exercises, concluding that there was no advantage in the use of these surfaces in closed chain exercises. Nevertheless, these authors did not differentiate between different types of unstable surfaces in their analysis. In an effort to deepen this topic, previous studies analyzed the EMG activity of the SA using different unstable surfaces types (Herrington, Waterman & Smith, 2015; Horsak et al., 2017). The authors did not observe significant differences in SA activity between the support surfaces, concluding that these unstable devices would increase the neuromuscular activity of the glenohumeral and trunk stabilizers muscles to control the center of mass displacement rather than increase scapular muscle stabilizers EMG activity.

Rehabilitation programs establish that in early stages it is necessary to increase SA activation above UT to improve movement and stability of the scapulothoracic joint and reduce the risk of subacromial impingement (Ludewig & Cook, 2000; Kibler et al., 2008; Youdas et al., 2018). In this way, the support of the upper limb on unstable devices, such as suspension equipment or therapeutic ball, would be recommended to improve the activation levels of the UT once an adequate muscular balance has been achieved between the scapular rotator muscles. Additionally, other factors such as the location of the unstable device may lead to a more favorable activation of the SA compared to the UT. In this context, a greater myoelectric activity of SA has been reported when the unstable device is located below the lower limb during the execution of closed kinetic chain exercises (Byrne et al., 2014; Youdas et al., 2020a); however, future research is necessary to corroborate this.

Conclusions

Different unstable support surfaces induce an increase in the myoelectric activity of the UT during a push-up performance, suggesting based on ES, a progressive neuromuscular demand, specifically higher EMG activity when using wobble board, therapeutic ball, and suspension equipment. Conversely, no differences were observed in SA, which could mean that performing a push up on unstable surfaces does not affect SA muscle activity level. However, these findings must be interpreted with caution due to reduced certainty of evidence for most outcomes. Nonetheless, current findings could be applied in the prescription of scapular muscle training exercises in healthy individuals and potentially in the rehabilitation of individuals with musculoskeletal dysfunctions of the shoulder complex, allowing to properly select the appropriate unstable device for a given stage of rehabilitation or training program.

Supplemental Information