Electromyographic Amplitude Ratio of Serratus Anterior and Upper Trapezius Muscles During Modified Push-Ups and Bench Press Exercises

Abstract

Imbalance and weakness of the serratus anterior and upper trapezius force couple have been described in patients with shoulder dysfunction. There is interest in identifying exercises that selectively activate these muscles and including it in rehabilitation protocols. This study aims to verify the UT/SA electromyographic (EMG) amplitude ratio, performed in different upper limb exercises and on two bases of support. Twelve healthy men were tested (average age = 22.8 ± 3.1 years), and surface EMG was recorded from the upper trapezius and serratus anterior using single differential surface electrodes. Volunteers performed isometric contractions over a stable base of support and on a Swiss ball during the wall push-up (WP), bench press (BP), and push-up (PU) exercises. All SEMG data are reported as a percentage of root mean square or integral of linear envelope from the maximal value obtained in one of three maximal voluntary contractions for each muscle studied. A linear mixed-effect model was performed to compare UT/SA ratio values. The WP, BP, and PU exercises showed UT/SA ratio mean ± SD values of 0.69 ± 0.72, 0.14 ± 0.12, and 0.39 ± 0.37 for stable surfaces, respectively, whereas for unstable surfaces, the values were 0.73 ± 0.67, 0.43 ± 0.39, and 0.32 ± 0.30. The results demonstrate that UT/SA ratio was influenced by the exercises and by the upper limb base of support. The practical application is to show that BP on a stable surface is the exercise preferred over WP and PU on either surfaces for serratus anterior muscle training in patients with imbalance between the UT/SA force couple or serratus anterior weakness.

Introduction

The serratus anterior and the upper trapezius are the main stabilizer muscles of the scapulothoracic joint (^13,16,21). These muscles act in a synergic action that allows for an appropriate scapulothoracic rhythm. This is essential to the maintenance of the length-tension relation of scapulohumeral muscles and to the normal biomechanics of the shoulder during humerus elevation movements (^13,20,22).

Previous studies have shown that in shoulder dysfunctions, such as glenohumeral instability (⁷), impingement syndrome (²²), and pain (²⁰), there is reduced electromyographic (EMG) activity of the serratus anterior muscle (^16,23,32). Excessive activity of the upper trapezius was observed in patients as a way to compensate for weakness of the anterior serratus, which may lead to a predisposition to impingement syndrome and shoulder pain. Thus, it is important to identify patients whose biomechanical dysfunction of scapulothoracic and glenohumeral joints results from muscle imbalance or weakness. The identification of imbalance of the force couple serratus anterior and upper trapezius has been represented through the UT/SA ratio proposed by Ludewig et al. (²²). The UT/SA ratio is the ratio of surface EMG amplitude between the upper trapezius and serratus anterior muscles. It is equally important to establish a rehabilitation program based on selective strengthening of the serratus anterior or the muscle considered the least active.

Studies concerning treatment or training of the upper extremity have analyzed EMG activation of the serratus anterior and upper trapezius, as well as that of other shoulder muscles during various closed kinetic chain (CKC) exercises (^{5,6,16,17,22,26}). CKC exercises are characterized by the application of axial loads and have been described as being favorable to the stimulation of proprioceptors, increased joint congruence, and greater capability for generating dynamic stabilization through muscle coactivation (^6,29). Hence, CKC exercises are recommended for promoting coactivation of muscle pairs that act on the shoulder and scapular girdle, such as the serratus anterior and upper trapezius.

Surface electromyography is frequently used to analyze the muscular demand required by the exercises. Most studies compare the EMG activity among different upper limb exercises that do not necessarily have biomechanical similarities, such as direction and load intensity, type of contraction, and extremity condition. However, previous studies have demonstrated that exercises with similar biomechanics produce EMG amplitudes comparable in primary mover muscle groups (^3,6). In addition, other factors may influence the EMG signal, such as changes in the relation speed and muscle length-tension, the type of muscle contraction (⁴), and, in cases of CKC exercises, the type of surface that supports the limb extremity: stable or unstable (¹⁷).

Rehabilitation programs have included exercises performed on an unstable surface. It has been suggested that training with this type of exercise reduces the prevalence of ankle injuries (³¹) and in cases of shoulder instability, improves joint proprioception (²⁷). However, little has been studied about the effects of using an unstable base of support on electric activity of upper extremity muscles (^1,17). Anderson and Behm (¹) demonstrate that, during the chest press exercise, an unstable surface was not capable of changing primary movers' EMG activity. This was shown despite the fact that the resulting joint torque was smaller because of the greater participation requirement of stabilizing and synergist muscles in joint stabilization. In their study, Behm et al. (²) examined isometric contractions of the quadriceps and plantar flexor muscles and observed a reduction in force production and in EMG activity in the recordings obtained during contractions performed on an unstable surface. Vera-Garcia et al. (³⁰), conversely, observed higher values of EMG amplitude for abdomen muscles during the curl-up exercise performed on an unstable surface. Lehman et al. (¹⁷) demonstrated that EMG amplitude of some muscles may be influenced by an unstable surface. For example, the triceps brachii presented higher muscular activity during the exercise performed on an unstable surface than when performed on a stable surface.

Thus, this study's purpose was to evaluate how the UT/SA ratio behaves in three different upper limb exercises (wall push-up, bench press, and push-up), aiming to verify whether it is influenced by the type of exercise, despite being biomechanically similar, and by the type of upper limb support surface. Previous studies demonstrate that primary movers present similar EMG amplitude among biomechanically comparable exercises (⁶). Other studies demonstrate that an unstable surface is capable of generating higher activity in stabilizing and synergist muscles (¹). Based on these two types of studies, in this study, we hypothesized that the proposed exercises would produce similar EMG values for the UT/SA ratio and that these muscles' EMG amplitude levels would be higher for the unstable surface.

Methods

Experimental Approach to the Problem

CKC exercises performed on a stable and an unstable surface are typical in rehabilitation and training programs (^5,22,27,31). In this study, we examined which surface and exercise elicited the greatest UT/SA ratio, suggesting a dominant activation of the serratus anterior associated with a lesser activity of upper trapezius muscle fibers. The EMG signals were collected with surface electrodes, processed with the root mean square (RMS) algorithm for stable surface and with integral of linear envelope for the unstable surface, and normalized by a maximal voluntary isometric contraction. A linear mixed-effect model was performed to compare UT/SA ratio values and to test the hypothesis that the proposed exercises would produce similar EMG values for the UT/SA ratio. An analysis of variance (ANOVA) was performed to test the hypothesis that studied muscles' EMG amplitude levels would be higher for the unstable surface. Exercises (wall push-up, bench press, push-up) and type of surface were randomized to reduce potential order threats to the study's internal validity.

Subjects

The subjects of this study were 20 men with the following characteristics: average age 22.8 ± 3.1 years, mean body weight 68.7 ± 7.9 kg, average height 1.75 ± 0.05 m. All subjects were right-handed, sedentary, and healthy. Volunteers were considered sedentary if they exercised less than three times a week, performing up to two different physical activities non-specific for the upper limbs and with no relation to sport training. Upper limb conditions were verified through history, inspection, palpation, and clinical tests (²⁴). Subjects were excluded from the study if they had a history of trauma of the scapular girdle or upper limbs, visible hypertrophy, trigger points in the evaluated muscles, and positive results from the clinical tests for impingement syndrome, lateral or medial epicondylitis, and joint instability of the shoulder, elbow, or wrist. Movement restriction in the shoulder joint and/or presence of painful arch of motion during arm elevation movements were also considered exclusion factors because they may cause variations in the EMG patterns in the various arm and scapular girdle muscles (¹⁵). Volunteers signed a consent form, according to norm 196/96 of the Brazilian National Health Council, approved by the ethics committee of the University Hospital (Hospital das Clínicas) of the Ribeirão Preto Medical School at the University of São Paulo.

Equipment

EMG data of serratus anterior and upper trapezius muscles were collected using single differential surface electrodes (two Ag bars, 10 × 1 × 1 mm, gain of 20, input impedance of 10 GΩ and common mode rejection ratio >80 dB) (EMG System do Brasil, São José dos Campos, São Paulo, Brazil). SENIAM recommendations (¹¹) were followed to position the electrode on the upper trapezius muscle, with the electrode placed on a line midway between the acromion and the seventh cervical vertebrae. For the serratus anterior, electrode positioning followed recommendations by Hintermeinster et al. (¹²), with the electrode fixed on the region under the axilla between the upper edge of the latissimus dorsi muscle and the lower edge of the pectoralis major.

In both muscles, electrode positioning was confirmed by manual muscle testing. A circular electrode (3 cm²) positioned on the sternal manubrium region was used as a reference electrode to reduce effects from electromagnetic interferences and other acquisition noises. Skin at the electrodes' sites was shaved, abraded, and cleaned with alcohol before the attachment of the electrodes to reduce skin impedance and obtain the best fixation. Recording electrodes were fixed using adhesive tape and positioned parallel to the direction of the muscle fibers and the reference electrode using conducting gel.

Surface EMG signals were collected by three channels of the Myosystem Br-1 electromyographer (Datahominis Tecnologia Ltda, Uberlândia, Minas Gerais, Brazil), of which two channels were used for the EMG and one for the load cell. The electromyographer has a reference device, simultaneous acquisition of all channels, digital filter pass-band of 20-500 Hz, input impedance of 10 GΩ in differential mode, 12-bit A/D converter with a 4-kHz sampling rate, and an amplitude range of −10 to +10 V. A load cell (Model MM; Kratos Dinamometros Ltda, São Paulo, São Paulo, Brazil) with nominal capacity of 100 Kgf was attached to the A/D converter and allowed force values to be recorded simultaneously with EMG signals. Audio feedback was used to inform volunteers about the produced force level, allowing the force to be maintained within a 10% variation of the established value. EMG and force signals were processed using Myosystem Br-1 2.9 software (Datahominis Tecnologia Ltda).

Testing Procedures

Three maximal voluntary isometric contractions (MVICs) of 6 seconds each were performed with the dominant limb to normalize the EMG data of the studied muscles. Ekstrom et al. (⁹) has suggested muscle testing positions that demonstrate higher EMG values of MVIC for the serratus anterior with the shoulder flexed at 125°, and for the trapezius with the shoulder abducted at 90°, the neck tilted to the same side and rotated to the opposite side, exerting an extension force. The maximal value of three MVICs served as a reference value for the normalization of integral of linear envelope value and RMS for each studied muscle on the unstable and stable surfaces, respectively.

After being evaluated and included in the study, volunteers performed three repetitions for each exercise with the purpose of learning the task and adapting to the audio feedback. Three isometric exercises were performed on a stable and an unstable surface: wall push-up, bench press, and push-up (Figure 1A-F). For the wall push-up, the subject remained in the orthostatic position, maintained the dominant upper limb flexed to 90° at the sagittal plane with the hand extended to 90° and elbow in full extension. The nondominant upper limb remained positioned at the side of the body, the trunk position was controlled to avoid rotations, and the distance from the wall was maintained at approximately the upper limb's length. An iron support with holes to attach the stable surface to the load cell was fixed on the wall and adjusted to the volunteer's height. For the bench press, the volunteer was positioned in dorsal decubitus, with feet supported on the stretcher, dominant upper limb flexed to 90°, wrist extended to 90°, elbow in full extension, and the nondominant upper limb stayed at the side of the body. An iron support fixed on a stretcher was also arranged to attach the stable surface to the load cell. Height was regulated according to that of the volunteer. The push-up was performed with the volunteer in the ventral decubitus position, with the thigh and knee flexed to 90°, trunk in neutral position (that is, parallel to the floor), and the dominant upper limb flexed to 90°, hand extended to 90°, and elbow in full extension. The nondominant upper limb was supported on the back of the body. A wooden box was placed under the subject's knee to ensure appropriate posture during the exercises, and a wooden support was arranged to attach the stable surface to the load cell.

Figure 1: Three isometric axial load exercises for the upper limb accomplished on a stable base of support and on a Swiss Ball: (A and B) wall push-up, (C and D) bench press, and (E and F) push-up.

On a stable surface, all exercises were performed with the hand of the dominant upper limb applying perpendicular force directly on this surface. On the unstable surface, a Swiss ball was positioned between the volunteer's hand and the stable surface.

Exercises were performed in a random sequence and repeated three times, each repetition lasting 6 seconds. Subjects rested for at least 1.5 minutes between isometric contractions and 3 minutes between each exercise to avoid fatigue effects in the EMG signal, which was collected simultaneously with the exertion.

Maximal isometric effort (100%) specific to each volunteer was determined for each exercise based on the average of three MVIC values recorded by the load cell in the three exercises, which lasted 6 seconds each. The force intensity applied by volunteers was controlled by audio feedback, which allowed for a maximal variation of 10% in relation to the pre-established force value. When recording the standardized MVICs, instructions and verbal commands during the exercises were always given by the same rater, who ensured that volunteers did not make any compensatory movements.

EMG Processing

EMG signal processing was done in 4-second windows (i.e., the first and last seconds of the 6-second recording were excluded for each studied task). EMG amplitude values are represented by the RMS for the stable surface and by integral of linear envelope for the unstable surface. These have been suggested among the possible presentation forms by the standardization norms for surface EMG studies (¹¹). The raw data collected from exercises accomplished on an unstable surface support were normalized by the ratio between the mean value of the integral of linear envelope obtained in each exercise and the maximal value of the integral of linear envelope obtained in three MVICs collected during manual muscular testing for each muscle. This procedure was also performed for the normalization of the raw data obtained from exercises accomplished on a stable surface, by the ratio between the mean RMS value obtained in each exercise and the maximal RMS value obtained in three MVICs. UT/SA ratio was calculated by the ratio of normalized values of the upper trapezius and by the normalized values of the anterior fibers of the serratus muscle, with RMS values used for the stable surface and the integral of linear envelope values for the unstable surface. The ratio's value was considered low if it was <0.3 (i.e., with an activation of the serratus anterior muscle three times greater in relation to the upper trapezius). The highest ratio values were close to or greater than 1, which indicates a similar activation of both muscles or that the upper trapezius muscle was dominant (²²).

Statistical Analyses

With the purpose of evaluating the influence of the surface and type of exercise on the average UT/SA ratio values and normalized EMG amplitudes of the studied muscles, a mixed linear effect model (²⁵) was used. This type of statistical analysis is recommended when the responses of one individual are grouped and the supposed independence among observations in the same group is not adequate (²⁸). In this case, the responses (UT/SA ratio values) may be considered as grouped by volunteers, and the information of each volunteer submitted to each studied exercise is used in the form of random effects. After building the model, residue analysis was performed, and the logarithmic transformation was adapted to meet some suppositions associated with the proposed model. This was done using SAS PROC MIXED, version 8 (¹⁹). An ANOVA was used to compare the muscles' EMG amplitude for each exercise and type of surface. A 5% level (P ≤ 0.05) was used to determine statistical significance.

The intraday and interday reliabilities of the UT/SA ratio values were calculated for the 20 volunteers by the intraclass correlation coefficient (ICC 2,1) (³³), obtaining excellent (¹⁰) intraday reliability (ICC 0.75-0.99), and interday reliabilities ranging between good and excellent (¹⁰) (ICC 0.58-0.87) on both surfaces.

Results

UT/SA ratio values are presented in Table 1. The result of applying the mixed linear effect model to evaluate the influence of surface on the ratio obtained during the studied exercises revealed that UT/SA ratio value was significantly higher (P < 0.0001) for the bench press performed on an unstable surface. No significant difference was found between UT/SA ratio values obtained in the wall push-up or push-up performed on stable and unstable surfaces. UT/SA ratio results for the comparison among the various exercises performed on the same surface showed statistically significant differences among all exercises performed on a stable surface (P < 0.0001) and only for the wall push-up and push-up performed on an unstable surface (P < 0.01) (Table 1).

Table 1: UT/SA ratio values for normalized electromyographic amplitudes

The normalized EMG activity of the serratus anterior was significantly lower for the wall push-up (P = 0.001) and bench press (P < 0.0001) performed on an unstable surface, whereas the upper trapezius muscle showed significantly reduced EMG activity only for the wall push-up (P = 0.03) performed on an unstable surface (Table 2).

Table 2: Comparison of the average normalized EMG amplitude values for corresponding muscles in each exercise on both bases of support.

Discussion

This study's results demonstrate that the UT/SA ratio value was influenced by the proposed axial load exercises, despite being biomechanically similar, and by the type of base of support for the upper limb. In the present study, we observed that controlling biomechanical aspects, such as load direction and intensity, contraction type, and extremity condition, was not enough to ensure similar UT/SA ratio values. The present results disagree with those of Dillman et al. (⁶), who evaluated the wall push-up, bench press, and push-up exercises in isotonic contraction and observed a comparable EMG activity for primary movers of the shoulder. Nevertheless, Dillman et al. (⁶) used equal load intensity for the compared exercises, whereas in the present study, the load required in isometric contractions was based on 100% maximal effort that volunteers were capable of performing in the standardized position for each evaluated exercise. For wall push-up and push-up exercises, the reaction force to maximal isometric effort against the surface, either stable or unstable, would cause rotation of the volunteer's trunk. This was controlled to maintain the standard position without additional support, as was the case for the bench press. This caused different force levels generated for the wall push-up, bench press, and push-up. The scapular stabilization in the bench press allowed individuals to exert greater isometric effort than in the other exercises and may also have allowed the serratus anterior a higher activation level as a primary mover, changing the UT/SA ratio value in relation to the other exercises.

Although a difference in the method can be observed in terms of the load applied in the various studies, this study's results indicate that more caution is needed when comparing exercises performed on a stable surface than those on an unstable surface. That is because the UT/SA ratio value varied between all the exercises on a stable surface (i.e., the wall push-up vs. bench press, wall push-up vs. push-up and between the bench press and push-up).

UT/SA ratios were different between the wall push-up and push-up on both surfaces. This difference may have resulted from rotational force acting on the upper limb during the wall push-up. That is because the upper limb weight force vector, which may be represented in the limb's center of mass during the wall push-up, is absent in the push-up, in which the weight force vector coincides with that of the axial compression force. Hence, our results agree with the suggestion of Lephart and Henry (¹⁸), who believed that exercises should not be classified based only on the extremity condition and the presence or absence of load. Rather, other factors should be considered, such as load direction, whether it is rotational or axial, and expected neuromuscular responses. A factor that may also have contributed to the different UT/SA ratio values between the wall push-up and the push-up is the incapability of subjects to perform maximal isometric effort in those exercises because of their position during the task.

Ludewig et al. (²²) analyzed UT/SA ratio values in four different variations of push-ups plus on a stable surface: standard push-up, knee push-up, elbow push-up, and wall push-up. UT/SA ratio values were considered low if they were less than 0.3, and those close to or greater than 1.0 were considered high. In the present study, the wall push-up was also responsible for the highest UT/SA ratio values compared with the other exercises, which is in accordance with the findings by Ludewig et al. (²²). However, UT/SA ratio values in the present study were 0.69 and 0.73 for stable and unstable surfaces, respectively, whereas those reported by Ludewig et al. (²²) were less than 0.5, considering the plus phase, which involves maximal scapular protraction.

In the literature, exercises involving protraction and/or scapular upward rotation are responsible for a higher level of EMG activity of the serratus anterior (^5,8,16,21,26). In their study, Ludewig et al. (²²) evaluated exercises with the plus phase, which are characterized by a maximal scapular protraction movement at the end of the exercise's concentric phase. Hence, it was expected that the exercises in the present study, which involved the scapula in a neutral position, would generate higher UT/SA ratio values than those reached by Ludewig et al. (²²).

In the present study, comparing one exercise performed on different surfaces resulted in a higher UT/SA ratio value only for the bench press on an unstable surface. This occurred as a result of two factors combined: a reduction in the serratus anterior muscle EMG amplitude and the maintenance of EMG activity by the upper trapezius in relation to the stable surface.

Exercises that showed low UT/SA ratio values (i.e., dominant activation of the serratus anterior associated with lesser activity of upper trapezius muscle fibers) may be an important component in the rehabilitation of patients with an imbalance between the activation of this force couple. Thus, considering the low ratio value, as well as a higher EMG activity level of the serratus anterior muscle, the bench press on a stable surface is preferred over the push-up and wall push-up in the rehabilitation program of patients who need to selectively activate the serratus anterior muscle.

Serratus anterior EMG activity levels observed in this study were reduced on an unstable surface in most exercises. This agrees with Behm et al. (²), who observed a reduction in EMG activity of quadriceps muscles and plantar flexor during isometric contractions performed in a sitting position on an unstable surface (Swiss ball). However, upper trapezius EMG amplitude levels remained constant for most exercises on both surfaces. This result is in agreement with Anderson and Behm (¹), who performed the chest press exercise, similar to the bench press, with the trunk supported on both an unstable (Swiss ball) and stable (bench) surface. It was then observed that EMG activity was maintained during isotonic contraction of trunk muscles and shoulder movers on both surfaces, whereas there was a reduction in force on the unstable surface compared with the stable surface.

The serratus anterior is the primary muscle responsible for scapular stabilization, whereas the trapezius acts as a supplementary stabilizer (^7,8). This may explain the greater influence that the unstable surface has on the serratus anterior, and the analysis of this muscle better represents the changes associated with instability. Thus, the results from this study are in agreement with those presented by Behm et al. (²).

The decrease in serratus anterior EMG activity in most of the exercises performed in the present study may have been caused by the higher recruitment of other shoulder muscles. Another reason may be that the high level of instability caused by performing the exercise unilaterally may have limited the individual's ability to activate and apply the force while maintaining balance. The synergetic participation of other shoulder muscles in scapular stabilization may have been performed by the lower trapezius, rhomboideus, and elevator scapular (¹³) muscles, which were not evaluated in the present study.

Results from the present study also revealed that the normalized levels of EMG activity generated in the studied exercises ranged from 5% to 47% of the amplitude value achieved during maximal voluntary contraction in manual muscle testing. Lehman et al. (¹⁷) also found normalized EMG amplitude values too low to produce increased muscular force. However, the authors attributed these findings to the fact that the studied samples consisted of subjects considered athletic and suggested that a population with less physical preparation could achieve activation levels capable of increasing muscular force. Results from the present study, which were based on the classification proposed by Kelly et al. (¹⁴), demonstrate a moderate level of muscular activation only for the serratus anterior in the bench press on a stable surface. This disagrees with suggestions by Lehman et al. (¹⁷) because the studied sample consisted of subjects considered sedentary (i.e., who performed low-intensity physical activities less than twice a week), and such activities were never directed to the upper limb, scapular girdle, or trunk.

Instability recruits shoulder muscles in a more general way and thus promotes greater stability and proprioception. This makes exercises on an unstable surface less harmful to the joint and possibly more advantageous than those performed on a stable surface. This is because they allow for different levels of muscular contraction to be achieved with less force (¹). Nevertheless, to generate EMG amplitude levels capable of strengthening musculature, the instability offered by exercises on an unstable surface must be moderate (²), preventing muscle action from being completely focused on joint stabilization. Regarding exercises on a stable surface, there is agreement that this kind of support is more appropriate when aiming to produce maximal force of primary movers (¹).

There is no consensus in terms of the influence that unstable surfaces have in increasing or reducing EMG amplitude. Nonetheless, it is generally observed that an unstable surface is capable of maintaining the EMG amplitude when the studied muscle performs the role of stabilizer and force generator proportional to the main function of generating force on a stable surface (¹). However, when the stabilizing force of the studied muscle can be synergetically performed by other muscles, the degree of participation of that synergist musculature, whether high or low, can determine the reduction or maintenance in the EMG amplitude, respectively, as observed in the present study. Finally, the increase in EMG amplitude is observed in muscles whose stabilizing function is not shared with other musculatures, as it is evidenced for the triceps braquial in the study by Lehman et al. (¹⁷).

It is important to take into consideration the difficulties in comparing previous studies with the results from the present study. For instance: 1) the type and velocity of contraction, as most studies involved dynamic contractions; 2) load intensity, usually fixed for all the studied exercises; 3) type of muscle, single or multi-jointed; 4) classification of exercises, open or closed kinetic chain; and 5) scarcity of studies concerning upper limb exercises performed on an unstable surface. Caution was taken in this study to control the variables that could have any influence on the EMG signal when performing exercises in isometric contraction and controlled axial load, which could limit the extrapolation of these data for the exercises performed dynamically. However, this procedure allowed for reliable data interpretation, revealing nuances in the use of a stable or unstable surface for common exercises in rehabilitation and training programs for the upper limb.

The fact that this study was performed with young healthy volunteers implies that further studies should be performed to evaluate the results to a population with shoulder dysfunction. However, the fact that the volunteers were sedentary, instead of athletes, may imply that patients with shoulder dysfunction would be able to perform the exercise easily. In addition, exercises were isometric, which are easier to perform, because they demands less effort from patients than isotonic exercises and do not involve painful arch.

There was, however, a limitation to the study because of the absence of analysis of other shoulder muscles' EMG activity. Such information could improve knowledge about how an exercise and its base of support influence EMG amplitude in CKC exercises for the upper limb.

Practical Applications

The present study's results demonstrate that the UT/SA ratio was influenced by specific exercises, although biomechanically similar, and by the base of support for the upper limb. This study has practical importance in that it shows bench press on a stable surface as the exercise preferred over wall push-up and push-up on either surface for serratus anterior muscle training in patients with an imbalance between the UT/SA force couple or serratus anterior weakness.

Acknowledgments

The authors would like to thank The Quantitative Methods Center- CEMEQ of the Ribeirão Preto School of Medicine for the statistical study of data and The Shoulder's Group of Laboratory of Analysis of Posture and Human Movement. This study was supported by The State of São Paulo Research Foundation (FAPESP) and The National Council for Scientific and Technological Development (CNPq).