Submaximal contractions can serve as a reliable technique for shoulder electromyography normalization

Abstract

There are a number of ways to normalize electromyographical data, the most common of which is using a maximal contraction as a reference. However, this technique is not always practical. The purpose of present study was to assess the reliability of an electromyographical data normalization technique using standardized submaximal contractions. Twenty healthy subjects (ten male, ten female) were used for testing, which was performed using both surface and fine-wire electromyography over two sessions at 15, 30, 45, and 60 percent of the day 1 maximum force. There was a mean of 5.3 days between test days. Data were compared between days, and the resulting ICC and standard error of the measurement values indicate varying levels of reliability at each submaximal percent. All heads of the deltoid, the upper trapezius and the serratus anterior demonstrated good reliability for at least one submaximal condition. The latissimus dorsi and supraspinatus demonstrated moderate reliability for at least one submaximal condition. Finally, the infraspinatus demonstrated poor reliability under all conditions. For situations in which MVC is impractical or anticipated to change, EMG amplitude normalization to one of these submaximal percentages appears to be a viable technique, at least for most muscles.

Introduction

Electromyography (EMG) is a common technique that involves using electrical activity as an indicator of muscle force production. However, the levels of electrical activity are variable and affected not only by muscle activity, but by hydration level, adiposity, surface preparation, sensor location, and a number of other factors that may be out of experimental control (Allison et al., 1998; Halaki and Ginn, 2012; Kelly et al., 1996). If data collection is being performed in one session, these factors remain relatively consistent, and data can be compared within each subject. However, in order to make comparisons between sessions or subjects, EMG data are generally normalized (Burden, 2010; Halaki and Ginn, 2012). Normalization can be performed in a number of ways: peak magnitude during a movement, average magnitude across a movement, magnitude while performing a reference contraction, or magnitude during an M-wave (Bolgla and Uhl, 2007; Halaki and Ginn, 2012; Yang and Winter, 1984). But the most common method of normalization is to use a maximal voluntary contraction (MVC) (Boettcher et al., 2008; Burden, 2010).

When research is performed using a patient population, and an MVC contraction could be painful and lead to further damage, or could be unrepresentative of a true maximum (Allison et al., 1998; Halaki and Ginn, 2012; Hsu et al., 2006; Norcross et al., 2010). Additionally, experimentations involving an intervention could affect an individual’s MVC, making it a suboptimal reference point (Park et al., 2008). Even without an intervention, maximal contractions have been shown to be more variable and less reliable than submaximal contractions (Dankaerts et al., 2004; Yang and Winter, 1984).

An alternative method is to normalize to a value at the same relative intensity and can be repeated between sessions and subjects. The aim of the present study is to investigate one such method, assessing the viability of normalization to an isometric contraction performed at a submaximal percentage of subjects’ MVC force. The goal is to assess between day reliability of submaximal isometric contractions when used for EMG normalization.

Methods

Subjects

Twenty subjects with an age range of 20-39 years (mean 25.0, SD 5.7 years) were recruited. Ten of the subjects were female, ten male; sixteen subjects reported right hand dominance, four left hand. Exclusion criteria were any history of surgery or rotator cuff tear in either shoulder, needle-induced syncope, pain while performing any of the prescribed movements or any neurological disorder. The University of Oregon Institutional Review Board approved the project and all subjects signed an informed consent form before participating.

Experimental set up

All instrumentation and data collection were conducted by a single investigator. The subject’s dominant side was instrumented with wireless Trigno EMG electrodes (Delsys Inc., Boston, MA) and wireless MTi inertia measurement unit (IMU) sensors (Xsens, The Netherlands). The skin on the dominant side was abraded and cleaned with alcohol, then insertion points for fine-wire EMG collection were determined for the supraspinatus and infraspinatus, based on standard guidelines (Delagi et al., 2011; Geiringer, 1999). Wires were inserted into the bellies of these two muscles and then connected to spring sensors for fine-wire EMG data collection (Basmajian and De Luca, 1985; Kelly et al., 1996). Surface EMG sensors were placed atop the anterior, middle, and posterior deltoid, upper trapezius, serratus anterior, and latissimus dorsi, based on standard guidelines (Cram et al., 1998).

Kinematic data were collected to allow for comparison at constant angles between days. One IMU was placed on the lateral aspect of the dominant arm, superior to the elbow. The other was placed on the contralateral torso at the level of the xiphoid process. Force was collected through a mounted load cell (Omega Engineering Inc., Norwalk, CT). All data were collected using custom Motion Monitor software at 2000 Hz (Innovative Sports Training, Inc., Chicago, IL). The EMG data were filtered at 20-450 Hz through the Delsys hardware. The Common Mode Rejection Ratio of the system was greater than 80 dB.

Maximal voluntary contractions

Two positions were used for isometric contractions. External rotation was used for the infraspinatus and latissimus dorsi and humeral elevation was used for the supraspinatus, upper trapezius, serratus anterior and all three heads of the deltoid. Humeral elevation was performed at 90 degrees of elevation and approximately 30 degrees forward of the frontal plane, elbow extended. External rotation was performed at approximately 10 degrees of abduction, elbow flexed to 90 degrees. The order of contractions (humeral elevation, external rotation) was randomized for each subject. The order was maintained for the second day of testing, which was performed between 2 and 9 (mean 5.3) days after the first day of testing. The MVCs were completed on both the dominant and non-dominant sides, with strong verbal encouragement provided. The subject performed each MVC for approximately three seconds and alternated between the dominant and non-dominant sides. Once all MVCs had been performed, they were performed again for a second attempt in the same order. The attempt with the highest force was used for analysis.

Submaximal and dynamic contractions

The MVCs performed on the first day of testing were used to determine the submaximal target forces for each subject. The order of submaximal contractions (15, 30, 45, 60 percent) was randomized and the order was maintained on the second day of testing. The subject was provided with a computer screen that displayed a visual range of +/− 2 percent around the target force, (e.g. 13-17 percent for the 15 percent contraction) and a line indicating the force currently being exerted. The subject was instructed to press against the load cell hard enough to get between the lines and to hold that force for approximately three seconds. Each submaximal contraction was performed twice in each position on the dominant side. Data were averaged across the two trials. Once the submaximal contractions were complete, subjects performed three repetitions of humeral elevation in the scapular plane.

Data analysis

A Root Mean Square (RMS) analysis was performed using a 200-millisecond window. For the maximal contractions, the peak force was determined and the 200-millisecond window around that peak was used. For the submaximal contractions, a range of plus/minus two percent of the target force (the same range given to the subjects for visual feedback) was calculated and the first 200 milliseconds inside that window were used to represent the EMG at that submaximal percentage.

IMU data were processed using the ISB standard for humerothoracic kinematics (Wu et al., 2005). These data were used to determine the point at which the subject achieved ninety degrees of humerothoracic elevation. A time window of 200 milliseconds around that point was selected as the portion of the EMG data to be normalized using each technique. Data were averaged across the three repetitions of humeral elevation motion.

To allow for comparisons between reference contractions, submaximal EMG values were divided by the effort level. For example, the value obtained at 30 percent was divided by 0.3 to calculate a projected value at 100 percent.

Statistical analysis

An intraclass correlation coefficient (ICC) was determined for each method of normalization (four submaximal percentages and MVC). The ICC model used was a two-way mixed effects, where people effects are random and measured effects are fixed (ICC 3,2). SPSS (IBM Corporation, Release 26.0.0.0) was used to perform these analyses. We adopted the following levels: poor reliability (< 0.5), moderate reliability (0.5 to 0.75), good reliability (0.75 to 0.9) and excellent reliability (>0.9) (Koo and Li, 2016). The standard error of the measurement was also calculated for each method of normalization. Additionally, a paired t-test was used to compare the forces produced by the dominant and non-dominant arms during MVCs.

Results

The reliability of the submaximal normalization methods varied across muscles and normalization levels (table 1, figure 1). All three heads of the deltoid demonstrated good reliability at 45 percent effort, each of which was higher than the MVC reliability. Beyond that, the posterior deltoid also demonstrated good reliability at 30 percent effort and the anterior deltoid demonstrated good reliability at all efforts except 15 percent. The upper trapezius demonstrated good reliability at 15 percent effort and this value was very similar to the MVC reliability. The serratus anterior demonstrated good reliability at 30, 45 and 60 percent effort, all of which were higher than MVC reliability.

Table 1.

EMG Amplitude Normalization ICCs and SEMs

	Normalization technique
	15%		30%		45%		60%		MVC
Muscle	ICC	SEM (%)	ICC	SEM (%)	ICC	SEM (%)	ICC	SEM (%)	ICC	SEM (%)
Anterior Deltoid	0.70	3.2	0.77	5.4	0.85	4.7	0.87	4.8	0.75	9.5
Middle Deltoid	0.51	3.8	0.71	4.9	0.81	3.6	0.70	5.7	0.68	7.7
Posterior Deltoid	0.57	3.7	0.79	3.7	0.85	3.2	0.58	6.4	0.70	5.0
Upper Trapezius	0.81	4.5	0.74	6.0	0.63	8.7	0.63	7.0	0.82	5.4
Latissimus Dorsi	0.06	12.3	0.51	11.5	0.45	16.7	0.47	17.2	0.77	8.3
Serratus Anterior	0.61	3.6	0.83	4.1	0.80	4.0	0.78	5.9	0.73	6.2
Supraspinatus	0.41	5.9	0.65	6.9	0.45	9.6	0.68	8.1	0.22	3.5
Infraspinatus	0.13	10.7	0.24	14.2	0.23	11.1	−0.05	16.7	0.15	3.7

Figure 1.

Graphs of EMG Amplitude Normalization. EMG amplitude at ninety degrees of humeral elevation normalized to four submaximal (15, 30, 45, 60 percent) values and to MVC. Data represent means and standard errors of the means for the 8 muscles tested: a) Anterior Deltoid, b) Middle Deltoid, c) Posterior Deltoid, d) Upper Trapezius, e) Latissimus Dorsi, f) Serratus Anterior, g) Supraspinatus, h) Infraspinatus.

There were no submaximal contractions for the latissimus dorsi, supraspinatus or infraspinatus that demonstrated good reliability. However, the latissimus dorsi demonstrated moderate reliability at 30 percent and the supraspinatus demonstrate moderate reliability at 30 and 60 percent. For all cases, including MVC, the infraspinatus demonstrated poor reliability.

Finally, with respect to force production, there was no significant difference between sides for the MVCs during humeral elevation (p=0.85) and external rotation (p=0.83).

Discussion

The normalization of EMG relies upon the repeatability of the reference contraction. While referencing MVC is the most common normalization technique (Burden, 2010), studies have shown maximal contractions to be less reliable than submaximal contractions (Dankaerts et al., 2004; Netto and Burnett, 2006). A previous study in our lab has also shown that patients’ ability to perform a MVC is directly influenced by pain (Ettinger et al., 2016). Additionally, they could cause further damage or pain in a patient population (Allison et al., 1998; Halaki and Ginn, 2012; Hsu et al., 2006). Normalization to MVC also provides analytical challenges when the MVC changes (Halaki and Ginn, 2012; Yang and Winter, 1983). Some of these challenges can be managed by referencing a contraction that can be more readily repeated (Allison et al., 1998; Halaki and Ginn, 2012; Norcross et al., 2010).

To our knowledge, the present study is the first assessment of reliability of submaximal isometric contractions of shoulder muscles. We found that there were submaximal efforts levels that resulted in good reliability for all heads of the deltoid, upper trapezius and serratus anterior. Moderate reliability was found for the latissimus dorsi and supraspinatus. However, the infraspinatus only demonstrated poor reliability.

Few studies have investigated the use of a submaximal isometric contraction for normalization purposes. Of those that have, most have used either a limb held against gravity or while holding an absolute weight (Allison et al., 1998; Chapman et al., 2010; Dankaerts et al., 2004). Two studies have investigated a submaximal load being moved in an isokinetic (Netto and Burnett, 2006) or isotonic (Allison et al., 1993) manner. Three other studies have investigated the reliability of an isometric contraction at a set submaximal percentage (Burnett et al., 2007; Yang and Winter, 1983, 1984). Yang & Winter (Yang and Winter, 1983) first compared both 30 and 50 percent isometric contractions to MVCs and found both submaximal contractions to have higher reliability than the MVCs for surface EMG of the triceps brachii, the one muscle being monitored (Yang & Winter, 1983). Yang & Winter (1984) later compared a 50 percent isometric contraction with other surface EMG measures, not including MVC. They found an increase in inter-subject variability (relative to raw EMG values), without an assessment of reliability, in leg muscles. Burnett et al. (2007) compared 60 percent isometric contractions to MVCs and found that both were highly reliable, with submaximal contractions being no less reliable than MVCs.

This study was not without limitations. Only four submaximal percentages were assessed; another percentage might be more reliable. The EMG of the latissimus dorsi and serratus anterior were both collected using surface sensors; considering the extent to which these muscles slide beneath the skin, fine wire EMG might have been more appropriate. Lastly, the data analysis was performed with the assumption that the prescribed submaximal percent was achieved; the actual force might have been up to two percent higher or lower.

Conclusions

These results demonstrate that shoulder EMG normalization at a submaximal percentage may be a viable technique that has reliability similar to normalization to MVC, for most muscles. Additionally, the maximal force exerted by one shoulder can be used to calculate a target force for a submaximal contraction using the other shoulder. This technique could be useful when assessing muscle activity in cases of an acute injury or a chronic condition.