In this study, muscle activity of 130 shoulders was analysed during a 30° arm abduction and adduction movement with and without additional handheld weight. Although only 25 patients with unilateral rotator cuff tears were recruited, incidental MRI findings were discovered in 62 additional shoulders. These results allowed a realistic evaluation of the load-induced increase in activity in shoulder muscles in different rotator cuff pathologies.

In our abduction test, a load-induced increase in muscle activity was observed in all shoulder muscles and types studied, where every 1-kg load increment resulted in a significant increase in muscle activity. The only exception was the activity of the latissimus dorsi muscle in shoulders with rotator cuff tendinopathy, which was less affected by additional weights. This low activation of latissimus dorsi at maximum abduction angle could be explained by its function, which is mainly arm adduction and internal rotation [12]. A load-induced increase in deltoid and rotator cuff muscle activity was also observed by Alpert et al. [24] in healthy subjects during the first 90° of abduction in the scapular plane. Analogous to that study, we observed similar muscle activity of the anterior and middle deltoid and a slightly lower muscle activity of the posterior deltoid. This is related to the individual muscle force vectors and muscle moment arms of the three parts of the deltoid. Indeed, the posterior deltoid muscle has a short lever arm during the first phase of full abduction [40]; moreover, the loaded and unloaded test was performed with the arm in neutral rotation, so no increased muscle activity of the posterior deltoid was required. The load-induced increase in muscle activity of the major muscles producing abduction torque (deltoid) is required in healthy subjects during the first 90° of abduction in the scapular plane to execute the movement. A systematic increase with additional weights in upper trapezius muscle activity during arm abduction in the scapular plane was also observed in asymptomatic subjects by Reed et al. [25]. This load-induced increase in activity of the other shoulder muscles is biomechanically supported because greater activity of the deltoid muscle causes potential translational forces on the humerus [14, 40], possibly leading to subacromial impingement, such that the rotator cuff and axioscapular muscles require increased muscle activity to counterbalance these forces [41]. We also observed an increase in muscle activity of the pectoralis major and biceps brachii muscles, possibly to oppose the action of the deltoid muscle, hence acting as a humeral head depressor. With increasing load, we also noted increased variability in muscle activity, which could be due to the individual strength capacity of the participants, as the additional weights were not scaled to the relative maximum strength. However, even in this case, the variability would have tended to increase with increasing load [24, 25].

Overall, shoulders with rotator cuff tears or tendinopathy showed the same load-induced increase in muscle activity as healthy shoulders, but relative muscle activity was higher in these patients. Moreover, symptomatic rotator cuff tears had higher muscle activity than asymptomatic rotator cuff tears, and these in turn had higher muscle activity than rotator cuff tendinopathies. In the latter case, there was one exception: pectoralis major muscle activity was lower in asymptomatic rotator cuff tears than in rotator cuff tendinopathy. This higher muscle activity in shoulders with rotator cuff pathology is consistent with the study of Kelly et al. [18], which concluded that patients with rotator cuff tears tend to have higher muscle activity compared with normal subjects, regardless of the presence of pain or symptoms. However, in contrast to the study of Shinozaki et al. [23], which used positron emission tomography (PET), we found no decrease in deltoid muscle activity and no increase in trapezius muscle activity in symptomatic rotator cuff tears compared with asymptomatic rotator cuff tears. This is likely due to the different methodologies and acquisition time of PET and EMG, and differences in the movement performed (arm in internal rotation versus neutral rotation). Although differences in upper trapezius muscle activity between symptomatic and asymptomatic rotator cuffs were not significant (log-transformed data), upper trapezius muscle activity tended to be higher, suggesting less glenohumeral joint motion in symptomatic rotator cuff tears. Significant differences between symptomatic and asymptomatic rotator cuff tears were found only in posterior deltoid and pectoralis major muscle activity, where the activity was greater in the symptomatic shoulders. One possible explanation may be that tears are more severe in symptomatic shoulders, leading to even greater compensation of the greater deltoid for the deficient rotator cuff and higher pectoralis muscle activity to counterbalance the superior translational forces of the deltoid muscle.

In another study, patients with symptomatic rotator cuff tears were found to have higher posterior deltoid and biceps brachii muscle activity, especially during weight lifting, compared with age-matched healthy controls [19]. Consequently, this study supported the benefit of treating the long head of the biceps tendon in symptomatic rotator cuff tears, as the biceps brachii can act as humeral depressor and cause pain if over-activated. This was only partially confirmed by our results, because in symptomatic rotator cuff tears, muscle activity of the biceps brachii and posterior deltoid was increased compared with healthy shoulders, but also the activity of the other muscles studied was increased. The increase in muscle activity was highest in the posterior deltoid muscle. While in healthy shoulders the muscle activity of the posterior deltoid is much lower than that of the anterior and middle deltoids, in shoulders with rotator cuff tears the posterior deltoid is rather active and its activity level approaches that of the anterior and middle deltoids. Therefore, the posterior deltoid muscle gains importance in abduction movement up to 30° in the scapular plane after rotator cuff tears. The observed increase in latissimus dorsi muscle activity is consistent with the study by Hawkes et al. [21]. Indeed, an increase in muscle activity of the biceps brachii, upper trapezius–serratus anterior, latissimus dorsi and teres major muscles was observed in massive rotator cuff tears compared with healthy subjects [21], as a compensatory mechanism for the destabilising forces of the deltoid. Similar to symptomatic rotator cuff tears, activity differences in almost all muscles were found between asymptomatic rotator cuff tears and healthy shoulders, but not in the pectoralis major muscle activity.

Clinically, rotator cuff tendinopathy is not as relevant as a rotator cuff tear, yet athletes with rotator cuff tendinopathy had an abnormal pattern of scapular movement that may be related to scapular muscle deficits [42]. During the loaded and unloaded abduction test in our study, shoulders with rotator cuff tendinopathy had higher activity than healthy shoulders in almost all muscles studied at all loads, with the exception of the infraspinatus and deltoid muscles at only some loads. Although latissimus dorsi muscle activity was higher in shoulders with rotator cuff tendinopathy, it remained unchanged with additional load, and deltoid and infraspinatus muscles had reduced activity. It is possible that these changes occur to avoid an overactivation of the tendinopathic supraspinatus muscle. Although the changes in muscle activity in shoulders with rotator cuff tendinopathy may not be as pronounced as in rotator cuff tears, compensatory mechanisms for the pathologic rotator cuff still appears to occur. Understanding the adaptive changes in muscle activity is crucial for rehabilitation as shoulder and scapula muscle activity may be altered with specific interventions such as mobilisation and strengthening exercises [43,44,44].

The possibility of comparing muscle activity of shoulders with rotator cuff pathologies and healthy shoulders with a simple abduction test is useful in the clinic to gain a better understanding of compensation mechanisms. In this 30° arm abduction test in the scapular plane, not only the deltoid and infraspinatus muscle showed a significant increase in activity with additional loading in all shoulder types, but also the surrounding stabilising muscles. In the Constant score, symptomatic rotator cuff tears differed from all other shoulder types and differences between asymptomatic rotator cuff tears and healthy shoulders were also detected. These differences were also observed in the muscle activity in our 30° arm abduction test. In addition, differences in the muscle activity between shoulders with rotator cuff tendinopathy and shoulders with asymptomatic rotator cuff tears or healthy shoulders were also present. This low abduction angle allows even patients with a limited range of motion to perform this test, and objective measurements of these patients can be obtained. However, performing tests with the same absolute handheld weights in all participants leads to variability due to individual strength capacity that must be taken into account when interpreting results. It is possible that weaker participants require a higher level of muscle activity (closer to maximal contraction to complete the task) or cannot perform the test at all, while stronger participants might not be as challenged by the same handheld weight. The EMG system is portable and the test can be performed by a trained person in less than 15 min, making an implementation in clinical practice feasible.

A major strength of our study is the combination of EMG data and MRI to investigate differences in muscle activity and potential compensatory mechanisms in shoulders with rotator cuff pathologies. EMG surface electrodes were used in this study, and because of the compact anatomy of the shoulder, the possibility of crosstalk of EMG signals from adjacent muscles cannot be excluded. To compare muscle activity between participants, we normalised the EMG with MVC. However, MVC might be influenced by pain and result in a larger normalised value due to a smaller denominator [46, 47], making it challenging to compare symptomatic (painful) and pain-free shoulders [20]. Alternatively, the amplitude could have been normalised to the unloaded condition of each shoulder, but this would not have excluded evasive muscle activity a priori. We chose the MVC normalisation method because there were no obvious differences in millivolt values between patients’ shoulders compared with the controls (Additional file 5).

Muscle activity exhibits some variability, to which several factors may have contributed. Although verbal instructions were given to the participants to maintain a comparable movement velocity, there could have been variations in movement duration. In addition, movement in the scapular plane was not restricted, and hence slight deviation from the scapular plane may have occurred. Some of the variability in muscle activity may also be explained by the heterogeneity of rotator cuff tears in the participants. However, to further characterise muscle activity for a specific tear type and severity, a larger number of participants would be needed.

Nonetheless, the results of this study are clinically relevant: this 30° abduction test was implemented to investigate the effect of additional handheld weight in shoulders with rotator cuff pathologies, and we indeed observed important biomechanical changes in shoulders with rotator cuff pathologies, such as greater relative activation of shoulder muscles even with small additional load.