Investigation of the association between shoulder flexion range of motion and deltoid and trapezius muscle activity after reverse total shoulder arthroplasty

Abstract

[Purpose] We aimed to evaluate deltoid and trapezius muscle activity during shoulder flexion postoperatively in participants who had undergone reverse total shoulder arthroplasty (rTSA) and identify factors contributing to active flexion range of motion (ROM). [Participants and Methods] We included 18 participants who had undergone rTSA and were followed for at least one year postoperatively. Surface electromyography recorded muscle activity of the deltoid and trapezius at 30° of shoulder flexion, approximately 8 months after surgery. Muscle activity was normalized as % reference voluntary contraction (%RVC). Based on active flexion ROM, participants were categorized into a good group (≥120°) and a poor group (<120°). [Results] The good group included 14 participants (3 males, 11 females, mean age 75.4 ± 5.7 years), and the poor group included 4 participants (1 male, 3 females, mean age 74.5 ± 11.4 years). The good group showed greater anterior deltoid activity, while the poor group exhibited higher activity in the posterior deltoid and lower trapezius. [Conclusion] We identified anterior deltoid activity as a positive factor for achieving good shoulder flexion ROM after rTSA. In contrast, increased activity in the posterior deltoid and lower trapezius may reflect compensatory strategies that inhibit optimal flexion.

INTRODUCTION

Reverse total shoulder arthroplasty (rTSA) was developed by Grammont and has since been used commonly for several indications worldwide owing to improvements in implants and surgery^{1, 2)}. The most important feature of rTSA is that the humeral head is inverted towards the glenoid fossa and the center of rotation of the glenohumeral joint (GHJ) is shifted medially and inferiorly^{3, 4)}. Comparisons of the results before and after rTSA have reported good postoperative outcomes in terms of pain and range of motion (ROM)^{2, 5)}. In terms of biomechanics, the rTSA group showed greater scapular upward rotation during flexion and an increased proportion of scapulothoracic joint (STJ) movement than healthy shoulders^{6, 7)}.

However, ROM gain varies among patients, with some patients experiencing difficulty in gaining flexion ROM. Risk factors for poor flexion ROM after rTSA include age⁸⁾, posture⁹⁾, obesity¹⁰⁾, diagnosis^{11, 12)}, and glenosphere position^13,14,15). Clinically, patients with poor flexion ROM after rTSA often show excessive upward rotation of the scapula, such as the shrug movement, from the beginning of the flexion movement. Matsuki et al.¹⁶⁾ investigated the dynamics of the GHJ and STJ during shoulder flexion in patients who underwent rTSA using two-dimensional/three-dimensional registration methods and reported that poor flexion ROM after rTSA was characterized by smaller GHJ motions and larger STJ motions. Our hospital uses the Scapula-45° imaging method as a functional imaging method for evaluating the shoulder complex, including the GHJ and STJ¹⁷⁾. This method uses four plain radiographs in a combination of no load and 3 kg load in a dropped shoulder position and 45° elevation above the scapular plane. The method was originally developed as a simple method for evaluating rotator cuff and scapular function, and we have adopted it for rTSA cases to evaluate the GHJ and STJ in no-load tasks alone. Studies by Takahashi et al.¹⁸⁾ and Maeda et al.¹⁹⁾ on rTSA cases using the Scapula-45° imaging technique reported that scapular function showed a similar pattern before and after surgery in patients who underwent rTSA, and that increasing the amount of change in GHJ may lead to a better elevation angle in the poor rTSA group.

The deltoid muscle is reportedly the primary source of force for rTSA and that preoperative and postoperative deltoid strength and muscle activity were significantly correlated with postoperative flexion ROM^20,21,22). Therefore, deltoid dysfunction may have a significant impact on the postoperative ROM gain. In the previous work of Inoue et al.²³⁾ reported a significant positive correlation between preoperative anterior deltoid (AD) volume index and postoperative flexion ROM. In addition, some reports have indicated that the posterior deltoid (PD) and upper trapezius (UT) muscle activity was significantly higher in the rTSA group when compared to healthy shoulders during shoulder flexion^{21, 24)}, suggesting a difference in the ratio of deltoid and trapezius muscle activity between the rTSA shoulder and healthy shoulders during shoulder flexion.

However, most of the studies compared healthy shoulders with rTSA shoulders with good flexion ROM, and no study has compared deltoid and trapezius muscle activity in rTSA shoulders with good flexion ROM and in a poor rTSA group with difficulty achieving gain in shoulder flexion ROM. Since the poor rTSA group reportedly shows shrug-like compensatory movements from the beginning of flexion²⁵⁾, a difference in the deltoid muscle activity in the early phase of flexion is also expected.

This study aimed to evaluate how deltoid and trapezius muscle activity during shoulder flexion in postoperative rTSA patients is associated with the good and poor groups. Based on previous reports of Pelletier-Roy et al.²¹⁾ and Nakano et al.²⁴⁾, the good group defined as having an active flexion ROM of 120° or more, and the poor group of less than 120°. This study also identify the factors contributing to flexion ROM, including known predictors such as patient background factors, GHJ and STJ angles, simple postoperative scapula-45° radiographic imaging, and preoperative deltoid volume index. We hypothesized that the ratio in AD muscle activity and postoperative GHJ angle were factors significantly related to the good group.

PARTICIPANTS AND METHODS

A total of 42 patients underwent rTSA at the one facility between October 2020 and April 2023. Among them, 24 patients were excluded based on predefined eligibility criteria: proximal humerus fractures (n=5), revision (n=4), refusal to provide consent (n=10), and inability to complete a minimum follow-up of one year (n=5). As a result, 18 patients (4 males, 14 females; mean age, 75.2 ± 7.3 years) who met the inclusion criteria were enrolled in the study. All patients had at least one year of postoperative follow-up by April 2024 and were able to perform active shoulder flexion of at least 30°. The primary diseases were osteoarthritis of the shoulder in 10 patients, cuff tear arthropathy in 6 patients, and massive rotator cuff tear in 2 patients. The following surgical models had all onlay designs: Stryker’s Ascend flex (Portage, MI, USA) in 10 patients, Exactech’s Equinoxe (Gainesville, FL, USA) in 6 patients, and Zimmer–Biomet’s Comprehensive Reverse (Warsaw, IN, USA) in 2 patients. Surgery was performed by two senior surgeons (N, N or K, F). The rTSA was placed with a posterior humeral torsion angle of 20° and with the center pin of the base plate at the intersection of the long and short axes of the glenoid fossa. The surgical approach was an anterosuperior (AS) approach in 4 patients and a deltopectoral (DP) approach in 14 patients.

We explained the content of the study and the extent of data use to the participants in advance and obtained written and verbal informed consent in accordance with the Declaration of Helsinki. Approval was obtained from the Ethics Committee for Research Involving Human Subjects of the Graduate School of Health Sciences, Showa University (Approval No. 534).

For electromyography (EMG) measurements, the participant was first placed in a standing position with the shoulder joint in the mid-position of internal and external rotation, the elbow joint in extension, and the wrist joint in the mid-position. Muscle activity was then measured using surface EMG while the shoulder joint was held at 30° of flexion for 7 seconds. EMG activity was also measured for 7 seconds with the shoulder in the dropped position for normalization.

EMG analysis is commonly performed using the percentage of maximal voluntary contraction (MVC) of each muscle to the normalized muscle activity during each movement²⁶⁾. However, MVC measurement for rTSA cases carries risks such as acromial fracture. In addition, taking the MVC measurement position is challenging in the poor rTSA group, which raises the question of whether they can demonstrate muscle activity appropriately. Therefore, based on the studies by Li et al.²⁰⁾ and Yun and Lee et al.²⁷⁾ we decided to use a normalization method called percent reference voluntary contraction (%RVC). This method involves dividing the EMG potential at an arbitrary constant angle by the EMG potentials measured in a standing posture at rest at each time point. Since %RVC is an alternative measurement for patients in whom MVC measurement is difficult and %RVC has been used in previous studies^{20, 27)}, we considered it to have a certain level of reliability. In the present study, we set a constant shoulder flexion angle of 30° as the minimum angle for the two groups to perform the same measuring task and applied a torque.

The six muscles tested were the AD, middle deltoid (MD), PD, UT, middle trapezius (MT), and lower trapezius (LT). The placement of the surface electrodes was based on the description of Delagi and Perotto²⁸⁾. A Trunk Solution TS-MYO (Tokyo, Japan) was used to measure surface EMG at a sampling rate of 1,000 Hz. Surface electrodes were applied after the skin was shaved, and dirt and lipids were removed with alcohol, and contact resistance was reduced with SkinPure (Tokyo, Japan) from NIHON. The surface electrode used in this study is a dry electrode with a constant distance between electrodes, and electrical signals are received via Bluetooth on an Apple iPad (Cupertino, CA, USA), a dedicated iOS terminal, for analysis. Unlike wired electrodes, the dry electrode is considered to provide a stable measurement with less baseline shift and noise intrusion.

EMG analysis software is an Excel macro for TS-MYO analysis from Trunk Solution. The original waveform was filtered using bandpass filtering, which transmitted 50–450 Hz from the sensor, it was smoothed using root mean square. From the resulting 7-second EMG waveform, the average EMG (AEMG) amplitude was calculated using 1-second sampling data, excluding the 3-second periods before and after the 7-second period. Next, %RVC was calculated by dividing the AEMG amplitude with the shoulder flexed at 30° by the AEMG of each muscle in the dropped shoulder position (Fig. 1).

Fig. 1.

Modified method of calculating the %RVC value.

The average electromyography (AEMG) value is calculated from the sampling data for 1 second, excluding the 3 seconds before and after the 7 seconds 30°flexion. % reference voluntary contraction (%RVC) is calculated by dividing the AEMG at 30° flexion by the AEMG at dropped shoulder position

The primary outcome measure was %RVC during a 30° flexion hold for each muscle. Secondary outcomes were flexion ROM, Constant Score, visual analog scale (VAS) score, GHJ angle, STJ angle using postoperative Scapula-45° imaging, and preoperative deltoid volume index.

Scapula-45° radiography, which is a technique used in our hospital to evaluate the function of the shoulder complex, including the GHJ and STJ, using simple radiographs¹⁷⁾, can easily measure the angle and motion ratio of the GHJ and STJ during shoulder elevation. With the patients in the seated position, the humerus is midway between internal and external rotation, and images are taken at 45° elevation above the scapular plane without load, taking care not to compensate for lateral flexion of the trunk. Although it is desirable to obtain images with the shoulder joint in a 30° flexion position, the GHJ and STJ overlap in the same position, making it difficult to measure the angles accurately. On the other hand, a 45° elevation above the scapular plane can clearly visualize both joints. In addition, the joint capsule and ligaments around the shoulder joint are reportedly most relaxed in this position²⁹⁾, which is considered appropriate for evaluating GHJ and STJ movements associated with muscle imbalance, with less influence from the joint capsule and ligament mechanisms. In this study, we expected that there would be differences in the GHJ and STJ movements associated with shoulder flexion movements between the good and poor groups, and the GHJ and STJ angles measured by the Scapula-45 imaging technique was selected as a secondary outcome owing to its potential to assess movements of the same joints as an adjunct. The angles were measured on the image, with the line connecting the upper and lower edges of the glenosphere as the basic axis, the angle formed with the humeral long axis as the GHJ angle, and the angle formed with the vertical line as the STJ angle (Fig. 2).

Fig. 2.

Scapula-45°radiography imaging method.

Both glenohumeral joint (GHJ) and scapulothoracic joint (STJ) are based on a line connecting the upper and lower edges of the Glenosphere, with the former measuring the humeral long axis and the latter the angle created by the vertical line.

Preoperative deltoid muscle volume was measured as an objective measure of deltoid muscle function. The deltoid muscle was visualized in three-dimensional images from computed tomography (CT) scans using Fujifilm Synapse Vincent (Tokyo, Japan). The origin (AD: lateral one-third of the clavicle, MD: lateral margin of the acromion, PD: scapular spine), insertion (deltoid tuberosity), and direction of each fiber were identified and classified as AD, MD, and PD, respectively (Fig. 3). Measurement reliability was analyzed using intraclass correlation coefficients ICC(1,2) and ICC(2,1) in previous reports, and values of 0.8 or higher were obtained with high measurement reliability²³⁾.

Fig. 3.

Image of deltoid segmentation on 3D.

Electromyographic measurements and simple radiographs were obtained at a mean of 8.2 months (range 6–12 months) postoperatively, and the Constant Score, VAS, and flexion ROM were measured preoperatively and simultaneously as above. Preoperative CT was performed for deltoid volume index at a mean of 2.4 months (range 0.5–4) prior to surgery. Patient characteristics were obtained from preoperative medical records.

In statistical analysis, patients were divided into two groups, with the good group defined as having an active flexion ROM of 120° or more, and the poor group defined as having an active ROM of less than 120°, as described in the introduction^{21, 24)}. JMP^®pro17.0.0 statistical software (SAS Institute) was used for statistical analysis. With the above two groups as binary variables, univariate logistic regression analysis was used to evaluate the association between %RVC and flexion ROM for each muscle, and the odds ratios (OR) and 95% confidence intervals (CI) were calculated. For some variables, the Fisher’s exact probability tests were applied due to sample size limitations. Bonferroni correction was applied to the primary outcome to account for the effect of multiple comparisons. The adjusted significance level was set at p<0.008. Meanwhile, the secondary outcome was interpreted as exploratory items and no multiplicity adjustment was applied; and the significance level was set at p<0.05.

RESULTS

The good group comprised 14 patients with 14 shoulders, and the poor group comprised 4 patients with 4 shoulders. The patient characteristics of the two groups were as follows: age, sex, height, BMI, diagnosis, surgery side, surgical approach, surgical model, flexion ROM, VAS, and Constant Score (Table 1). The mean postoperative flexion ROM for the good and poor groups was 150 ± 15° and 80 ± 5°, respectively. Regarding the primary outcome, the %RVC for AD (OR=1.319, 95% CI: 1.053–2.008, p=0.007) showed a significantly positive association with the good group. In contrast, %RVC of PD (OR=0.030, 95% CI: 0.000–0.372, p≤0.001) and %RVC of LT (OR=0.218, 95% CI: 0.022–0.0653, p=0.001) showed significantly negative associations with the good group. No significant association was found for MD, UT, or MT. Concerning the secondary outcomes, postoperative GHJ angle (OR=1.164, 95% CI: 1.039–1.502, p=0.001) and preoperative AD volume (OR=2.607, 95% CI: 1.115–50.480, p=0.006) showed a positive association with the good group. In contrast, the STJ angle (OR=0.849, 95% CI: 0.678–0.955, p=0.002) showed a negative association with the good group (Table 2). Note that the postoperative flexion ROM and postoperative Constant Score were probably directly related to the dependent variable, making univariate logistic regression analysis challenging; the OR and 95% CI were listed as Not Applicable (NA) (Table 2).

Table 1. Patient characteristics.

	Good (N=14)	Poor (N=4)	Overall (N=18)
Age (years)	75.4 ± 5.7	74.5 ± 11.4	75.2 ± 7.3
Sex (male/female)	3/11	1/3	4/14
Height (cm)	154 ± 7.5	152 ± 2.9	153 ± 7.1
Body mass index	20.1 ± 4.2	18.7 ± 2.4	19.9 ± 3.9
Diagnosis
Cuff tear arthropathy	4	2	6
Massive rotator cuff tear	1	1	2
Osteoarthritis	9	1	10
Surgery side(right/left)	8/6	2/2	10/8
Surgery approach
Anterosuperior	3	1	4
Deltopectoral	11	3	14
Surgery model (Onlay)	14	4	18
Flexion ROM
Preoperative (°)	70 ± 40	40 ± 30	60 ± 40
Postoperative (°)	150 ± 15	80 ± 5	130 ± 30
Visual analog scale
Preoperative	6.1 ± 2.1	6.0 ± 1.8	6.1 ± 1.8
Postoperative	2.1 ± 1.6	2.3 ± 0.8	2.2 ± 1.5
Constant score
Preoperative	30.6 ± 10.8	25.5 ± 2.5	29.5 ± 9.8
Postoperative	61.6 ± 7.0	35.8 ± 2.0	55.8 ± 12.7

ROM: range of motion.

Table 2. Logistic regression analysis of muscle activity and clinical outcomes.

Primary outcome
Each muscle	OR (Odds ratio)	95% CI (Confidence interval)	p-value
Anterior deltoid	1.319	1.053–2.008	0.007*
Middle deltoid	0.748	0.488–1.022	0.070
Posterior deltoid	0.030	0.000–0.372	<0.001**
Upper trapezius	0.884	0.686–1.128	0.307
Middle trapezius	0.398	0.063–2.129	0.272
Lower trapezius	0.218	0.022–0.653	0.001*
Secondary Outcome
Flexion ROM	OR (Odds ratio)	95% CI (Confidence interval)	p-value
Preoperative	1.023	0.990–1.072	0.179
Postoperative	NA	NA	<0.001**
Constant score
Preoperative	1.089	0.946–1.383	0.280
Postoperative	NA	NA	<0.001**
Visual analog scale
Preoperative	1.021	0.539–1.920	0.945
Postoperative	0.954	0.461–2.265	0.900
Surgery approach	1.222	0.050–14.683	0.880
Postoperative GHJ (°)	1.164	1.039–1.502	0.001*
Postoperative STJ (°)	0.849	0.678–0.955	0.002*
Preoperative deltoid volume
Anterior deltoid	2.607	1.115–50.480	0.006*
Middle deltoid	1.006	0.844–1.181	0.931
Posterior deltoid	0.865	0.700–1.012	0.071

Bold indicates statistically significant values. *p<0.01 **p<0.001

OR: odds ration; ROM: range of motion; GHJ: glenohumeral joint; STJ: scapulothoracic joint.

DISCUSSION

The results of this study showed no obvious differences in the demographic data between the two groups. %RVC of AD showed a significant positive association with the good group; postoperative GHJ angle and preoperative AD volume index showed a positive association with the good group. In contrast, %RVC of PD and LT showed a significantly negative association with the good group, and the postoperative STJ angle showed a negative association with the good group.

The primary active muscle during shoulder flexion is the AD³⁰⁾. Schwartz et al.³¹⁾ reported that the moment arm of the AD was significantly increased after rTSA, while the absence of the AD significantly decreased shoulder flexion, emphasizing the importance of the AD in rTSA. Li et al.²⁰⁾ also measured deltoid muscle activity before rTSA using surface EMG and reported a significant positive correlation between preoperative AD activity and postoperative flexion ROM. In the previous work of Inoue et al.²³⁾ reported a significant positive correlation between preoperative AD volume index and postoperative flexion ROM. The results of this study showed that the %RVC of AD and preoperative AD volume index showed a positive association with the good group, supporting the findings of previous reports and suggesting that the preoperative AD volume index may also affect AD activity in postoperative shoulder flexion movements. The results of postoperative GHJ angle were similar to those reported by Matsuki et al.¹⁶⁾ and Takahashi et al.¹⁸⁾, suggesting that increasing the GHJ angle in the early phase of shoulder flexion may be important for improving flexion ROM.

In contrast, PD and LT %RVC and postoperative STJ angle showed a negative association with the good group. PD is primarily responsible for shoulder extension and external rotation and is the antagonist muscle in shoulder flexion movements³⁰⁾. Mechanically, the activity of antagonist muscles is not necessary in shoulder flexion, and the movement can be performed without the activity of these muscles³²⁾. However, in the actual movement, the activity of the antagonist muscles (simultaneous contraction) is to increase joint stability, and the moment of the main active muscle is inhibited by the moment of the antagonist muscle³³⁾. Thus, in rTSA shoulder flexion, the main active muscle is the AD, and excessive activity of the PD, which is the antagonist muscle, may inhibit the moment of the main active muscle. This study suggests that the higher the %RVC of PD, the less likely it is to be classified as a good group, and considered the possibility that excessive muscle activity in PD resulted in the difficulty in shoulder flexion.

However, Nakano et al.²⁴⁾ measured the muscle activity of the periscapular muscles during shoulder flexion movement in healthy shoulders and rTSA cases (active flexion ROM >120° or more) and reported that the ratio of muscle activity of the PD was significantly higher in the rTSA group. Werthel et al.³⁴⁾ also examined the differences in the moment arm of the deltoid muscle before and after rTSA using a cadaveric shoulder and found that the extension moment of the PD was significantly smaller after rTSA, which may support flexion and abduction movements of the shoulder joint. In other words, the overactivity of the PD in the shoulder flexion movement of the poor group may have complemented the weak muscle activity of the AD and formed a compensatory movement. After rTSA, the traditional role of the deltoid as the primary active or antagonist muscle changes, and the PD may not inhibit shoulder flexion⁴⁾ Nevertheless, the AD are the primary active muscles in shoulder flexion after rTSA, and it is important to promote the activity of this muscle and inhibit the overactivity of the PD to improve flexion ROM.

LT is primarily responsible for scapular subduction, adduction, and upward rotation, and is considered to contribute to GHJ stability by rotating the scapula upward in cooperation with the serratus anterior muscle during shoulder joint flexion^{35, 36)}. This study showed that the higher the postoperative STJ angle, the less likely the patient was to be in the good group, and the higher the postoperative GHJ angle, the more likely the patient was to be in the good group. rTSA patients have been reported to have a higher ratio of UT to PD muscle activity during shoulder flexion than individuals with healthy shoulders^{21, 24)}; however, these reports are based on patients with good flexion ROM (active flexion ROM >120° or more). No study has investigated the muscle activity of the periscapular muscles in cases of poor flexion ROM as in the present study, suggesting that the contribution ratio of LT to muscle activity may be an important feature of poor flexion ROM in patients who have undergone rTSA.

This study had several limitations. First, the number of patients was insufficient, especially that in the poor group. The variation in the OR and CI resulting from the use of univariate logistic regression analysis may be attributed to the insufficient sample size. A post-hoc power analysis was conducted with an assumed power (1-beta error probability) of 0.8, an effect size of 0.3, and a significance level (alpha error probability) of 0.05, indicating that a total sample size of 88 participants would have been required to achieve adequate statistical power. However, the implant design in this study was all onlay, and the diseases included in the study were only those caused by chronic or degenerative disease^{11, 37)}. There was no obvious bias in the patient characteristics, and significant variables were detected in the Bonferroni-corrected p-values of primary outcome, believing that the influence of the composition of the inclusion cases was minimized as much as possible. Second, the motion analysis of GHJ and STJ was performed two-dimensionally alone using simple radiographs, and the motion was different from the actual measurement motion. Compared with 3D motion analysis, the Scapula-45° imaging method has the advantage of simply measuring GHJ and STJ motion, but it cannot consider horizontal and sagittal plane motion; therefore, the measurement was considered a secondary outcome of the good and poor groups in this study. Third, a combination of surgical approaches was employed. In particular, the AS approach is more invasive to the deltoid than the DP approach, and a systematic review by Seok et al.³⁸⁾ reported that flexion ROM was significantly higher with the use of the DP approach. In this study, 4 patients underwent the AS approach, 3 of whom were in the good group. Thus, the influence on the results may have been small; however, the effect of the different approaches on muscle activity cannot be excluded. Regarding diagnosis, Keller et al.¹¹⁾ reported that patients with proximal humerus fractures had significantly lower postoperative flexion ROM than those with rotator cuff arthropathy. Based on this, we excluded patients with traumatic conditions and included only those with chronic or degenerative diseases. Meanwhile, Saini et al.³⁹⁾ found that patients with OA showed better flexion ROM than those with CTA, suggesting that the presence and quality of residual rotator cuff tissue may affect functional outcomes. Although some studies suggest that differences in underlying diagnoses have minimal impact on postoperative outcomes^{40, 41)}, future research may benefit from including only a specific disease and assessing rotator cuff integrity. Beyond surgical approach and diagnosis, implant design and positioning may also affect outcomes. Haidamous et al.⁴²⁾ found that a larger glenosphere, greater inferior overhang, and posterior humeral offset improved forward flexion after rTSA, while lateralization and distalization showed no benefit. In our study, all implants were of the onlay type but varied by manufacturer, with no standardization in size or placement. Given conflicting evidence on implant positioning and ROM, our lack of detailed radiographic analysis may be a limitation. Future studies should control for implant variability and include quantitative evaluation of component alignment. Lastly, this study measured muscle activity and ROM at a mean of 8.2 months postoperatively (range 6–12 months); however, the final improvement of ROM and muscle activity after rTSA may require more than 2 years postoperatively⁵⁾. However, ROM improvement is reportedly greatest at 1 to 6 months postoperatively, with slow improvement at 6 months to 2 years postoperatively⁵⁾. Hao et al.⁴³⁾ also reported that when clinical outcomes were poor at 3 and 6 months after rTSA, 85% of these patients did not show significant improvement at 2 years postoperatively. Thus, the measurement period in this study could be considered as the period of symptom fixation. However, this study may not reflect the final improvement status of rTSA, and the possibility that some cases, especially those in the poor group, were in the process of recovery should be considered. Further studies should increase the sample size, introduce three-dimensional motion analysis, and conduct long-term follow-up studies to allow for more accurate analysis and interpretation.

Conflict of interest

The authors have no conflicts of interest directly relevant to the content of this article.