In this study, scapular and glenohumeral kinematics of 130 shoulders with and without rotator cuff disorders were examined during a bilateral 30° loaded and unloaded abduction test in the scapular plane. Most of the incidental findings on MRI occurred in participants over 45 years of age. Most of the asymptomatic shoulders had partial tears, while more than half of the symptomatic shoulders had at least a full-thickness supraspinatus tear (kinematics were not significantly different; Supplementary Table S3). Using single-plane fluoroscopy, we were able to evaluate both upward–downward scapular rotation and superior–inferior glenohumeral translation during a 30° loaded (up to 4 kg) and unloaded abduction test in shoulders with rotator cuff disorders and in healthy shoulders. Load-induced changes in scapular rotation were assessed in healthy shoulders and in shoulders with rotator cuff tears, while glenohumeral translation was unaffected by the additional handheld weights. Overall, the kinematic changes that occurred during 30° abduction, such as upward scapular rotation and superior glenohumeral translation, were mainly restored during 30° adduction with downward scapula rotation and inferior glenohumeral translation.

In the unloaded condition, upward scapular rotation during the 30° abduction movement increased to 2.3° in healthy shoulders and to 3.6° in shoulders with symptomatic rotator cuff tears. A similar increase in upward scapular rotation during abduction was reported in a suprascapular nerve block study [43]. In addition, Miura et al. [13] reported – albeit with slightly smaller values – an average increase in upward scapular rotation of 0.1° in healthy shoulders and of 3.3° in shoulders with massive rotator cuff tears. Their study used an electromagnetic tracking system, which might underestimate scapular motion owing to skin motion artefacts. In our study, kinematic differences were observed in scapular motion between pathologic and healthy shoulders, with shoulders with symptomatic rotator cuff tears having the greatest upward scapular rotation and healthy shoulders having the least upward scapular rotation during the 30° abduction movement. Some differences in scapular rotation were found between shoulders with rotator cuff tears (asymptomatic or symptomatic) and healthy shoulders, and also between shoulders with symptomatic rotator cuff tears and shoulders with rotator cuff tendinopathy at some loads. However, no major differences were found between shoulders with symptomatic and asymptomatic rotator cuff tears up to 30° of abduction. The presence of symptoms may thus be related to different compensatory muscle activities. Indeed, in the same umbrella study [32, 44], higher muscle activities in the posterior deltoid and pectoralis major were found between symptomatic and asymptomatic rotator cuff tears.

The finding of greater upward scapular rotation after rotator cuff tears is consistent with other kinematics studies [12,13,14,15,16,17] and is further supported by electromyography studies [45, 46], which have observed increased axioscapular (upper trapezius, serratus anterior) muscle activity as an indication of increased upward scapular rotation. Thus, the increased scapular rotation in shoulders with rotator cuff disorders may reflect a compensatory mechanism for the deficient rotator cuff muscles to counteract the decreased abduction torque.

Regarding glenohumeral translation, we observed a superior glenohumeral translation of 1.0 and 1.2 mm during the 30° abduction movement without additional weight in healthy shoulders and in shoulders with symptomatic rotator cuff tears, respectively. Similar values were reported by Kozono et al. [24] for 30° of abduction in the scapular plane, and hence, these small glenohumeral translations may simply be physiological. Although there were no significant differences between shoulder types in our study, shoulders with symptomatic rotator cuff tears may tend to have slightly larger superior glenohumeral translations because of the missing centralising action of the rotator cuff and greater activity in the deltoid muscle, as also supported by Kijima et al. [20]. Moreover, rotator cuff tears involving the infraspinatus tendon are likely to be associated with larger glenohumeral translations [47].

We observed a load-induced increase in scapular rotation in healthy shoulders and in shoulders with rotator cuff tears. Glenohumeral translation changed with additional weight neither in healthy shoulders, as reported also by Nishinaka et al. [30], nor in shoulders with rotator cuff tears during 30° abduction. Therefore, in this cohort, shoulders with rotator cuff tears and healthy shoulders appear to behave similarly under load, with upward scapular rotation increasing during 30° abduction so as not to alter glenohumeral joint stability. Thus, maintaining regular glenohumeral motion may be more important than maintaining regular scapular motion during abduction initiation. However, this behaviour may change at larger abduction angles, when the surrounding muscles have reached their maximum level of activation, possibly leading to greater glenohumeral translation. From a muscular point of view, it is possible that the additional load results in increased activity of the deltoid and the intact rotator cuff muscles, which are then counterbalanced by increased activity of the axiohumeral muscles (pectoralis major, latissimus dorsi), so that the axioscapular muscles (trapezius, serratus anterior) also become more active to perform the movement [44].

Shoulders with rotator cuff tendinopathy showed a different trend with additional loads during 30° of abduction, as a load-induced increase in superior (abduction) or inferior (adduction) glenohumeral translation was observed, while scapular rotation remained constant, possibly as a result of the electromyographic changes observed in the larger umbrella study, such as increased deltoid muscle activity and almost unchanged latissimus dorsi muscle activity with additional load [44]. In particular, the smaller load-induced increase in the latissimus dorsi muscle activity in shoulders with rotator cuff tendinopathy compared with the other shoulders may be related to greater glenohumeral translation. Hence the superiorly directed force of the deltoid (and supraspinatus) muscles is less counterbalanced by the adductor activity of the latissimus dorsi. However, the load-induced changes were small with a mean absolute change of 0.9 mm during 30° abduction with 4-kg load compared with 0-kg load. Depending on the load and the extent of the rotator cuff pathology, the level of compensation required may be greater or smaller, resulting in different scapular rotation.

The abduction test using single-plane fluoroscopy, as used in our study, could be implemented in clinical practice owing to the efficient image acquisition and processing to obtain accurate shoulder kinematics, which may help in deciding the treatment strategy. This method allows obtaining accurate glenohumeral kinematics and assessing compensatory strategies under different loading conditions that could be addressed with therapy [48] without exposing patients to excessive radiation. The absence of a strong association between the results of the abduction test and the Constant score suggests that these two tools may assess different aspects of shoulder functionality with rotator cuff tears. Overall, lower Constant scores are to be expected in symptomatic rotator cuff tears, but the abduction test might perceive subtle variations in the joint function. A greater instability might progressively wear down the joint and lead to secondary damage, such as osteoarthritis [49]. An accurate kinematics investigation may thus contribute to an in-depth assessment of the shoulder joint. Although only a 30° abduction–adduction movement was analysed in this study, greater upward scapular rotation was found in shoulders with rotator cuff disorders, and additional load also increased upward scapular rotation during 30° abduction. This suggests that the scapulohumeral rhythm is likely to decrease in the presence of rotator cuff disorders [12] or with load, which may be important to consider in clinical decision-making and physiotherapy training.

A major strength of our study is the assessment of shoulder kinematics on fluoroscopic images combined with MRI findings to better investigate and objectively quantify potential compensatory mechanisms in shoulders with rotator cuff disorders, as these may be asymptomatic. As a result, the types of tears in shoulders with asymptomatic and symptomatic rotator cuff tears were not homogeneous, as all shoulders with asymptomatic rotator cuff tears were incidental findings. Because of this regrouping, we were not able to account for side dependency in the analyses between both shoulders (e.g. affected side of patients and contralateral side). In addition, we were unable to assess the effect of age on shoulder kinematics owing to the large number of incidental findings, resulting in few healthy shoulders within the older group. The kinematic results show some variability to which several aspects may have contributed. In this study, we acquired images with single-plane fluoroscopy, and scapular rotation and glenohumeral translation were calculated using anatomical landmarks on the scapula and humerus. Kinematic accuracy would presumably have been higher with dual-plane fluoroscopy and the use of a 3D-to-2D model-to-image registration technique to track the scapula and humerus. However, the availability of dual-plane fluoroscopy in a clinical setting is limited, and bone models derived from computed tomography are required, which together would expose patients to high doses of radiation [50]. Single-plane fluoroscopy provides sufficient accuracy for in-plane motion [51], as in our case with image acquisition and arm movement in the scapular plane, and patients are exposed to low radiation doses. Because of the single-plane fluoroscopy, scapular rotation and glenohumeral translation could only be measured in one direction (upward–downward and superior–inferior, respectively).

Some of the variability in the data might be due to slightly different movement velocity, although verbal instructions were given during testing, or to small movement deviation from the scapular plane, since movement was not restricted. The extent and heterogeneity of the rotator cuff disorders among participants might also be responsible for some of the variability in scapular and glenohumeral kinematics. However, more study participants would be needed to specifically analyse a particular type of rotator cuff tear and its severity. Because of the small number of massive tears in this study, results of this study might not be generalizable to such a group. In addition, the handheld weights were not selected according to the individual strength capacity of each participant. This may have resulted in some variability in glenohumeral kinematics, as the relative effort to perform the test may have differed between participants, resulting in different levels of shoulder muscle activation. Nonetheless, the results of this study are clinically relevant. This 30° arm abduction and adduction movement with additional weights allowed the effect of load on shoulder kinematics to be studied in shoulders with rotator cuff disorders and in healthy shoulders. This test provides objective measures of shoulder instability, and relevant biomechanical changes were observed.