Corticosteroid injection combined with a shoulder rehabilitation system for chronic rotator cuff injuries: a protocol for a randomised controlled trial

Abstract

Introduction

The main clinical symptoms of patients with chronic rotator cuff injuries (CRCI) include pain and/or limitation of movement, which severely reduce the function of the shoulder joint. Local injection of corticosteroid with local anaesthetics can control the inflammatory response and effectively relieve patients’ pain in the short term. In addition, rehabilitation exercises are considered an important tool for improving shoulder function. However, due to the presence of pain or substandard execution of movements, it is often difficult for patients to achieve the desired therapeutic effect. Although there are many options for digital rehabilitation, relatively few purely conservative treatments have been used for patients with CRCI. And even fewer studies have addressed how to improve exercise accuracy in such patients.

Methods and analysis

This protocol comprises three phases. First, shoulder active motion data will be collected from patients with CRCI and analysed using K-means clustering to define distinct clinical rehabilitation stages based on movement patterns and biomechanical principles. The clinical stages will be validated using one-way analysis of variance (ANOVA) and cross-tabulation to evaluate interstage functional differences and clinical consistency. Second, a shoulder rehabilitation system with real-time feedback will be developed, and its usability evaluated through a pilot study incorporating the System Usability Scale, Simulator Sickness Questionnaire and semistructured interviews. Finally, a randomised controlled trial will be conducted. 60 participants will be randomly allocated to either G1 (corticosteroid injection+digital rehabilitation) or G2 (corticosteroid injection+traditional home rehabilitation). The primary outcome is the Constant-Murley Score. Secondary outcomes include range of motion, Numerical Rating Scales, University of California at Los Angeles Shoulder Score, movement accuracy and exercise attitude questionnaire. Between-group comparisons will use independent t-tests or Mann-Whitney U tests, and within-group changes will be analysed with repeated measures ANOVA or Friedman test, with post hoc Bonferroni-adjusted comparisons.

Ethics and dissemination

The protocol was approved by the Medical Ethics Committee of Gongli Hospital, Pudong New Area, Shanghai (number: GLYY1s2024-031). All participants will provide informed consent prior to enrolment. The study findings will be disseminated through publication in a peer-reviewed journal or presentations at relevant national and international academic conferences.

Trial registration number

ChiCTR2500097903.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study was technologically innovative and used virtual reality and motion capture technology for rehabilitation exercises.

• Only researcher 1 will be aware of the subgroups, while the other researchers and all participating patients will be blinded.

• This protocol provides a highly detailed and reproducible methodological framework, systematically covering shoulder motion pattern classification, digital rehabilitation system development with pilot testing and a preliminary clinical efficacy evaluation of combined injection and exercise therapy.

• The study proposes a multifaceted kinematic scoring system to quantitatively assess exercise standardness, including trunk inclination, target joint angle accuracy, compensatory motions and postural alignment during movement.

• The sample size is limited to a preliminary cohort and may be underpowered for subgroup analyses or rare adverse events. Additionally, the short-term follow-up period does not allow for assessment of long-term treatment effects or sustainability of outcomes.

Introduction

The use of the upper limbs is indispensable in the daily life of human beings, which is not only the guarantee of the basic needs of life but also the key to achieving a higher level of social function. As the connecting hub between the upper limb and the trunk, the shoulder joint is one of the most flexible and mobile joints in the body, capable of performing a variety of complex movements, such as combing one’s hair, touching one’s back and playing badminton, and so on. It is due to its high-frequency use¹ and limited subacromial space that the rotator cuff tissues (especially the supraspinatus tendon) are at an extremely high risk of injury. Studies have shown that the prevalence of rotator cuff injuries (RCI) in the general population is quite high, reaching 5.5%.² In addition, the process of strain is often caused by overuse and is not easily detected, and the blood supply to the rotator cuff tendons is relatively poor. These factors make it difficult for the rotator cuff tissue to repair itself after injury. As a result, RCI tends to be more often chronic. Patients may experience only mild discomfort in the early stages until the symptoms become severe enough to warrant attention.

After an injury, the rotator cuff usually develops a local aseptic inflammation. The main symptom is pain, which is particularly noticeable when the shoulder is in motion or at rest at night.³ A combined corticosteroid and local anaesthetic injection can provide effective pain relief by blocking the pain signal from the lesion to the central nervous system. To date, it has been widely demonstrated that this method has a significant analgesic effect.⁴ Moreover, modulating the inflammatory response creates a microenvironment more conducive to tendon regeneration, which can ultimately accelerate the recovery process in chronic RCI (CRCI).⁵

Shoulder pain after an injury causes the patient to reduce active movement of the shoulder joint to avoid worsening pain. The prolonged build-up of inflammatory factors and the patient’s maintenance of pain-avoiding postures for long periods may cause adhesions and contractures in the muscles and soft tissues around the shoulder. This can create a negative cycle that not only further limits the range of motion (ROM) of the shoulder joint but can also exacerbate pain. Rehabilitation exercises are a key component in restoring joint mobility,⁶ which not only effectively reduces shoulder pain but also significantly promotes the recovery of joint function. The compliance and accuracy of exercise are important factors that affect the effectiveness of exercise.^{7 8} During rehabilitation exercise, patients often lose confidence in continuing to exercise because of pain.⁹ In addition, to relieve pain or in an attempt to achieve a higher joint angle, patients may also unconsciously perform compensatory movements during exercise, which can interfere with the recovery process.

Digital healthcare systems—including augmented reality, virtual reality (VR) and motion capture technology—have been used in previous studies to rehabilitate patients suffering from RCI. Compared with traditional treatments, patients using digital therapies have demonstrated more significant improvements in the recovery of shoulder function.^{10 11} One study used an inertial motion tracker in conjunction with a mobile app for an 8-week exercise intervention for patients with shoulder pain. This approach demonstrated high comparability to traditional supervised face-to-face rehabilitation sessions. In both rehabilitation modalities, the patient group experienced improved shoulder function and a significant reduction in pain.¹² Additionally, some studies have also demonstrated the superiority of digital rehabilitation methods in reducing pain and discomfort, further proving the effectiveness of digital therapies in treating RCI.^{11 13}

Currently, there are various forms of digital shoulder rehabilitation programmes, including commercial games and various types of motion capture and guidance systems, which have facilitated the rehabilitation process to some extent. Rotator cuff pathology can be classified by severity; the Neer classification is a common system that includes stage I (oedema and inflammation), stage II (fibrosis and partial-thickness tears) and stage III (full-thickness tears).¹⁴ For patients with partial-thickness tears, a conservative approach incorporating corticosteroid injection and guided exercise is generally recommended as the initial management strategy.15,17 However, few studies have explored conservative programmes for patients with CRCI, particularly with regard to the timely correction of movements and the judgement of the motion’s accuracy during rehabilitation. It is noteworthy that patients identified with full-thickness tears or those who do not respond to conservative treatment will be referred for surgical evaluation to ensure timely intervention.^{18 19}

Given this background, the aim of this clinical trial study is to explore a conservative treatment option for patients with CRCI, which combines injection therapy with digital rehabilitation (performing exercises guided by a digital healthcare system), and to assess in depth the impact of the digital therapy system on shoulder function. The specific study objectives are as follows:

Objective 1: To analyse shoulder joint kinematic characteristics in patients with partial-thickness rotator cuff tears using K-means clustering, and to derive clinical rehabilitation stages based on movement patterns and biomechanical principles for establishing short-term exercise goals.

Objective 2: To develop a digital healthcare system based on VR and motion capture technology for functional exercise in CRCI and to gather patient feedback to inform its refinement.

Objective 3: To comprehensively evaluate the specific performance of two protocols, injection therapy combined with traditional home rehabilitation and injection therapy combined with digital rehabilitation, in relieving pain, improving shoulder function, expanding ROM and enhancing movement accuracy in patients with CRCI, so as to provide an effective protocol for patients who seek conservative treatment.

Method

Research design

This is an experimental research protocol of the type of randomised controlled clinical trial approved by the Ethics and Research Committee of Shanghai Gongli Hospital (number GLYY1s2024-031). First, a shoulder exercise system with VR and motion capture-based technology will be created based on the clinical symptoms of CRCI patients. A pilot study will then be conducted with a small cohort of participants to assess the system’s usability and identify potential improvements through standardised questionnaires and interviews, with subsequent optimisation of the system based on the feedback obtained. The researchers will then recruit subjects to participate in the trial and randomly divide them into two groups: G1 (local injection combined with digital rehabilitation) and G2 (local injection combined with traditional home rehabilitation). The overall flow of the protocol is shown in figure 1. And it will be conducted from June 2024 to February 2026 at the Department of Pain Management of Gongli Hospital. The study will report following the recommendations of the Standard Protocol Items: Recommendations for International Trials (online supplemental appendix A).²⁰

Figure 1. Protocol flow chart. VR, virtual reality.

Data collection and motion level definition: building a foundation for rehabilitation of RCI (objective 1)

Participants

In order to ensure that the sample size is sufficiently large to support the reliability of the analysis results, it is usually recommended that the sample size should be at least 10 times the number of variables during exploratory analyses. Given that the k-means cluster analysis method will be used in this study, this method requires a sample size of at least 100 cases to ensure the robustness and validity of the analysis. Therefore, in order to conduct an in-depth cluster analysis, data needs to be collected from at least 100 eligible CRCI patients, covering the mobility information of the four key directions: flexion, abduction, internal rotation and external rotation.

Inclusion criteria

To be recruited into this study, participants had to meet the following criteria:

• Men and women with an imaging diagnosis (MRI) of RCI²¹ and a non-total rotator cuff tear injury.

• Patients with a unilateral shoulder injury with pain for ≥3 months.

• Pain level of ≥3 on the Numerical Rating Scales (NRS).

• Age >18 years.

• Degenerative rotator cuff pathology.

Exclusion criteria

• Full-thickness tear of the rotator cuff, or the presence of a history of shoulder trauma.

• With other conditions that would result in shoulder dysfunction (such as shoulder fractures, labral injuries, osteoarthritis and shoulder dislocation).

• A history of previous surgery on the affected shoulder.

• A history of allergy to medications relevant to this study.

• Minors, women who are pregnant and breastfeeding, and patients who cannot understand or perform the exercise.

Operational rules

Active ROM data for shoulder flexion, abduction, external rotation and internal rotation will be measured in the standing position using a standard goniometer. Incomplete datasets will be excluded from subsequent analysis. The elbow method will first be employed to determine the optimal number of clusters. Subsequently, k-means clustering will be applied to classify the complete datasets into distinct groups. These data-driven groups will then be interpreted and defined as distinct rehabilitation stages based on anatomical and kinematic principles.22,25 A key operational rule will be applied: for each stage, the short-term functional goal is defined as the upper limit of the ROM for that stage. This approach ensures that goals are sequential, immediately achievable and personalised to the patient’s current capacity, thereby facilitating motivation and reducing the risk of reinjury.

System design and improvements: technical support for the implementation of rehabilitation programmes (objective 2)

Design of system content

Researchers and engineers work together to complete the intelligent shoulder rehabilitation system. This system is divided into two parts, which are the VR education subsystem and the motion capture subsystem. The researcher writes the scientific content of RCI through professional knowledge, which introduces the definition of RCI, aetiology, symptoms, diagnosis and examination, as well as treatment and rehabilitation methods. In addition, the researcher will provide the operating procedures of the motion capture subsystem to the engineer and cooperate with the engineer to shoot a video of standard movements as the scoring criteria for the accuracy of the shoulder rehabilitation system. The engineer will model, design and build the system according to the protocol provided by the researcher. The motion capture subsystem will first conduct a comprehensive assessment of the patient’s maximal ROM. The patient will then be automatically assigned to the exercise stage that corresponds to their current maximal ROM, as defined by the staging criteria established in objective 1. The subsystem not only guides the patient through muscle-controlled repetitions but also assesses the accuracy of the patient’s movements. It provides visual feedback to help patients adjust their movements in a timely manner.

Accuracy scoring criteria

Kinect camera is used to capture the patient’s movement data during training in real-time, which in turn is comprehensively assessed in four core dimensions. These include the degree of inclination of the patient’s torso, the angle of the shoulder joint in the specified direction of motion, the compensatory situation of the shoulder joint in the non-specified direction of motion, and whether a standard position is achieved during training (eg, the starting stance in internal and external rotation training). In addition, based on the degree of importance of each dimension, they will be assigned a corresponding proportion of weight, constituting an evaluation system with a score of 100 points. A higher score means that the patient’s movement is closer to the standard.

Pilot Usability Study

After the development of the rehabilitation system is completed, we will recruit subjects to experience the system, aiming to collect feedback and suggestions from a wide range of user groups through standardised questionnaires (eg, the System Usability Scale (SUS),²⁶ Simulator Sickness Questionnaire (SSQ)²⁷ and qualitative interviews. Subjects will include doctors and therapists, healthy individuals and patients with CRCI to ensure the comprehensiveness and representativeness of the evaluation. The sample size will be determined according to the principle of information saturation. This means that subjects will be recruited continuously until the analysis of the interview data shows that there are no significant new recommendations from the additional interview data.^{28 29} We will conduct a systematic analysis of all data collected to distill key feedback, including functional improvement points and optimisation suggestions, for targeted optimisation and improvement of the system.

Clinical trials and impact assessments: validating the effectiveness of rehabilitation systems

Sample size calculation

Minimum sample size calculations will be performed using G*Power V.3.1.9.7 software (http://www.gpower.hhu.de/en.html) prior to patient recruitment to improve the accuracy of the study. Based on previous studies, the Constant-Murley Shoulder Function Score (CMS) was used as the primary outcome measure in this study, taking the difference value between the two groups to be 5.1 and the SD to be 5.2.³⁰ 22 patients were inferred to be needed in each group when taking α=0.05, β=0.1, and the allocation ratio was 1. Considering a 20% dropout rate, 60 patients were eventually considered for inclusion in the trial.

Team of researchers

Four researchers will be involved in this study: Researcher 1 will be responsible for the randomised allocation of patients and will exercise and assess the patients using the Smart Shoulder Rehabilitation System; Researcher 2 will assess the patients’ general information as well as trial evaluation indicators and will instruct the patients on how to carry out the subsequent autonomous exercises using the VR education subsystem and Researcher 3 will carry out the ultrasound-guided injections for each patient; Researcher 4 will process and analyse the collected data. Since there is no subjectivity in the scoring of patients by the intelligent rehabilitation system, it will be permissible to be operated and used by researcher 1.

Randomisation

Researcher 1 used EXCEL to generate random numbers to determine group numbers. The numbers were then placed individually in sealed, opaque envelopes, which were drawn by the patients themselves to determine which group they would be placed in for the trial.

Blinding

In order to ensure the objectivity and impartiality of the findings, this study was designed to implement a series of rigorous blinding and randomisation instruments, aiming to effectively circumvent subjective bias in the evaluation of patients by the researchers and to diminish the possible expected effects of specific interventions by the patients. With the exception of researcher 1, all researchers and patients were blinded to the rehabilitation regimen they received. Patients’ names and unique numbers were recorded in detail as soon as they were randomly selected from the sealed envelopes, and the patients’ identities were anonymised by this number in all subsequent study sessions as a way of further protecting patient privacy and ensuring anonymity of data processing. The training of the patients in group G1, as well as the scoring of the accuracy of movement of all patients at the last follow-up visit, will be carried out by researcher 1. Patients are required to follow the prompts of the system. Researcher 2 will be responsible for thoroughly assessing and recording the baseline information of the patients, and the functional status of the shoulder joint. Besides, researcher 2 needs to instruct each patient on the correct self-exercise method by VR education subsystem, without knowledge of the group to which the patient belonged. Researcher 3 only precisely injects the injured and inflamed areas of the patients under ultrasound guidance. Finally, researcher 4 will receive the data in a thoroughly anonymised form and only have to process and analyse the data. The brief flow of blind assignment is shown in figure 2. When the patient goes out of the group or voluntarily requests a change in his treatment plan, his blind state will be revealed.

Figure 2. Blind method process.

Interventions

Ultrasound-guided corticosteroid Injection

Participants will receive an ultrasound-guided injection of a mixture of compound betamethasone and lidocaine. The procedure will be performed by a qualified physician with over 10 years of experience in musculoskeletal ultrasound and injection techniques. During the procedure, the patient will be placed in a supine position with the affected shoulder fully exposed. The physician will use a high-frequency linear array ultrasound probe to identify key anatomical landmarks, including the subacromial-subdeltoid bursa, the supraspinatus tendon and the acromion. After confirming the target location, the skin over the injection site will be disinfected thoroughly. Using an aseptic technique, the injection will be administered via an in-plane approach under continuous ultrasound guidance to ensure accurate needle placement and deposition of the injectate within the bursal space.

Digital rehabilitation

To avoid the problem of dizziness that may be caused by the VR glasses, each patient uses the VR education subsystem in the seated position to learn about the disease. The VR will introduce the function of the rotator cuff, the symptoms and causes of RCI as well as the methods of rehabilitative exercises to the patients. Specific exercises are as follows:

Flexion and abduction training: The patient holds a booster stick with both hands to perform forward flexion or abduction of the shoulder joint. When the affected shoulder joint reaches the highest angle of restriction, the healthy hand will move the affected side to the highest place that can be tolerated by the pain, and then slowly put it down after keeping it for 15–30 s.

Internal and external rotation training: The healthy arm presses down on the affected humeral head, while the affected shoulder is kept at 90° of abduction and 90° of elbow flexion. Patients are asked to train like maneki-neko in the maximum ROM. The above movements should be performed for 10–15 min every day, and no exercise should be performed on the day of injection treatment.

Motion accuracy scores for this system

Patients will be assessed twice before training and the maximum angle is taken to assign them to the appropriate phase for training. Shoulder injury is closely associated with proprioceptive deficiencies.³¹ Injury triggers inflammation, oedema and potential damage to proprioceptors, leading to abnormal sensory information input, decreased joint stability and an elevated risk of reinjury.³² Consequently, enhancing proprioceptive input is vital for shoulder rehabilitation. In this study, patients will be asked to start neuromuscular control training from the lowest angle of the training phase to the highest. This kind of training can increase proprioceptive input and promote recovery of shoulder function.³³ Whenever the patient reaches the current target angle, the next angle will be increased by 10° for training until the maximum angle of the shoulder joint is reached; if the patient does not reach the target angle, he or she will continue to use that angle as the training target again; if the patient fails to reach the target angle twice in a row, the angle will be reduced by 10°. The participants in G1 will take the motion capture subsystem for four directions of motion, with 15 training sessions in each direction, at the end of which they will be reassessed.

Assessments

Assessments will be made at four moments throughout the study: before treatment (T0), 1 week of treatment (T1), 1 month of treatment (T2) and 3 months of treatment (T3). All personal data as well as the results of the study will be stored in a digitally coded database. Researcher 1 needs to record the patient’s number and name at the patient’s first visit. Researcher 2 is responsible for recording the patient’s general information at baseline, including gender, age, side of injury, injured muscle and duration of disease. The assessment will then be measured at four time points from T0 to T3. At T3, Researcher 1 will also score the patient’s motion accuracy using the Smart Shoulder System. The assessment instruments to be used for primary and secondary outcomes and the chronological order of the evaluations are all shown in table 1.

Table 1. Indicators and chronological order of evaluation.

Measuring instruments	T0	T1	T2	T3
Primary
CMS	✔	✔	✔	✔
Secondary
NRS	✔	✔	✔	✔
UCLA	✔	✔	✔	✔
ROM	✔	✔	✔	✔
Exercise Attitude Questionnaire	✔	✔	✔	✔
Motion accuracy				✔

CMS, Constant-Murley Score; NRS, Numerical Rating Scales; ROM, range of motion; T0, before treatment; T1, 1 week of treatment; T2, 1 month of treatment; T3, 3 months of treatment; UCLA, University of California at Los Angeles Shoulder Score.

Primary outcome

The primary outcome indicator of this study focused on the assessment of shoulder function, using the widely recognised CMS,³⁴ which comprehensively covers four dimensions: pain perception, ability to perform activities of daily living, active ROM and abductor strength. The scale consists of 10 items with a total score of 100, of which 35 are subjective and 65 are objective, and the increase in score directly reflects the improvement of the patient’s shoulder function. In the objective assessment, the measurement of ROM is particularly critical, covering two items measuring the angle of flexion and abduction, as well as two items that are scored based on the completion of movement scoring, namely internal rotation and external rotation. It is worth mentioning that the CMS scale has been rigorously tested for reliability in the country where the research team is based and can be applied to this study.³⁵

Secondary outcomes

Patients’ pain levels will be assessed by the NRS scale, which consists of a total of 11 numbers from 0 to 10 indicating the severity of the pain. A score of 0 indicates no pain, with increasing numbers indicating increasing pain, and a score of 10 indicates that the patient is suffering from the worst pain and is unable to fall asleep.³⁶ Each number representing the degree of pain will be explained to patients prior to the assessment, and the corresponding number will be chosen by patients to express the level of pain as they perform specific activities.

In this study, the active ROM should be measured in all four directions of the patient’s shoulder joints: flexion, abduction, external rotation and internal rotation. Patients will be asked to perform the maximum active movement within the pain-free range in the standing position, while the physiotherapist will be required to measure the patients’ angle at the limit of movement using a protractor.

Another scale used to assess shoulder function will also be used in this study, the UCLA scale. It has a total score of 35, with higher scores representing better shoulder function. compared with the CMS scale, UCLA is more concise. It has only one assessment item per dimension and also assesses patient satisfaction with treatment. This item is scored out of 5 and will not be scored at baseline.

Accuracy scores will be given by the intelligent rehabilitation system. A higher score indicates a more standardised movement, but the score is independent of the angle.

Patients will also be expected to complete the Exercise Attitude Questionnaire from T0 to T4. The questionnaire is designed to collect information on patients’ attitudes towards exercise and the number of rehabilitation exercises completed. Attitude towards exercise consists of 10 questions about both positive and negative attitudes and one open-ended question ‘Why I do not want to do functional exercises’.

Statistical analysis

The statistical analysis will be performed using IBM SPSS Statistics (V.22.0) for quantitative data and NVivo (Release 1.2) for qualitative data analysis. For all quantitative outcomes, the normality of data distribution will be ascertained using the Shapiro-Wilk test, and the homogeneity of variances will be assessed using Levene’s test. Data conforming to normality will be expressed as mean±SD, while non-normal data will be expressed as median and IR (P25, P75). The statistical significance level will be set at p<0.05.

For kinematic data classification in objective 1, K-means clustering will identify movement patterns, with the optimal cluster number determined by the elbow method. The validation of rehabilitation stages will employ one-way analysis of variance (ANOVA) (or Kruskal-Wallis test) to compare functional scores across clusters, while cross-tabulation with Cohen’s kappa will assess classification agreement.

In objective 2, thematic analysis using NVivo will analyse qualitative interview data, while SUS and SSQ data will be analysed descriptively without formal hypothesis testing.

For the randomised trial in objective 3, categorical variables (eg, gender) will be compared using χ² tests. Between-group comparisons of outcomes (ROM, NRS, CMS, UCLA, Accuracy) will be performed using independent samples t-tests for normally distributed continuous variables or Mann-Whitney U tests for non-normal distributions at each assessment time point. Within-group changes over time will be assessed using repeated measures ANOVA for normally distributed data or Friedman test for non-normal data, with post hoc pairwise comparisons using paired t-tests or Wilcoxon signed-rank tests with Bonferroni correction.

Confidentiality

Researchers must inform patients participating in the study of the entire treatment process either verbally or in written text. Patients are also required to sign an informed consent document before they are enrolled in the study (online supplemental appendix B). This document should state the benefits and potential harms of the trial and the alternatives. Patients choose to voluntarily participate in the trial after reading and understanding the process. They can withdraw at any time without discrimination or retaliation, and any medical treatment and rights will not be affected as a result.

Ethics and dissemination

This study was approved by the Medical Ethics Committee of Gongli Hospital, Pudong New Area, Shanghai, China on 25 June 2024 (Approval Number: GLYY1s2024-031). It was filed in the National Health Security Information Platform on 8 January 2025 (Registration Number: MR-31-25-000935). Furthermore, the protocol was successfully synchronised with the Chinese Clinical Trial Registry on 27 February 2025 (Registration Number: ChiCTR2500097903). The results obtained from the subsequent implementation of this protocol will be published in a peer-reviewed journal or presentations at relevant national and international academic conferences, providing a conservative treatment method for RCI to clinicians, patients and researchers, and offering a treatment approach for other shoulder conditions.

Timeline

Figure 3 illustrates the timeline of this protocol. All activities are carried out according to schedule. The clinical trial phase will be completed by December 2025.

Figure 3. Timeline of this study.

Discussion

RCI, as a common shoulder disorder, can cause pain and dysfunction that seriously affects patients’ quality of life. Traditionally, local anaesthetics have been widely used for acute pain management due to their ability to effectively block pain signals, while hormones have been used to reduce the inflammatory response due to their anti-inflammatory effects. The combination of the two does provide significant short-term relief of shoulder pain in patients with RCI, which is consistent with previous studies.^{37 38} However, the long-term effects of the combination were not satisfactory. The pain was prone to recur after relief from injection therapy, and the functional improvement of the shoulder joint was not significant. This suggests that reliance on medication alone may not be sufficient to fully address the problems associated with RCI and that a more comprehensive treatment strategy needs to be sought.

Exercise-based rehabilitation methods are considered the first-line treatment for rotator cuff tendinopathy.⁶ In 2020, the Agency for Healthcare Research and Quality stated that exercise interventions are effective in relieving pain perception and improving physical function.³⁹ This includes, but is not limited to, building muscle strength, improving joint stability and promoting tissue repair. In addition, exercise can alleviate pain and significantly improve various related outcomes, including physical function, mood, fatigue and sleep quality.⁴⁰ Rehabilitation exercises are flexible. They can be performed by patients in different scenarios and can be seen as a continuation of hospital care after patients are discharged from hospital. However, the effectiveness of exercise performed by patients at home is hampered by multiple factors, particularly decreased motivation to exercise due to pain and inappropriate exercise and posture. Many patients are often reluctant to initiate exercise because of their fear of pain or lack of supervision. Furthermore, if patients perform movements improperly during rehabilitation, they may exacerbate the degree of shoulder injury. If there is excessive reliance on trunk compensation during exercise, although it appears that the mobility of the shoulder joint has increased, in reality, the function of the shoulder joint has not been effectively improved. All these factors will lead to the treatment effect not meeting their expectation, which seriously affects the patient’s confidence in the treatment.

Currently, digital therapeutics is experiencing robust growth and development. It can reduce pain intensity by effectively distracting the patient’s attention through an immersive environment.⁴¹ In view of this, this study proposes a new treatment protocol that combines both injection therapy and rehabilitation exercises and incorporates a Smart Shoulder Rehabilitation System as an assistive tool. This protocol is a randomised and blinded clinical trial type of experimental study designed to evaluate the effects of different rehabilitation exercise modalities combined with injection therapy on pain and shoulder function in patients with CRCI. The system designed in this study can not only instruct patients on how to perform rehabilitation exercises but also monitor and score the patient’s movements in real-time with sensors and algorithms to ensure that the exercises are standardised and effective. In addition, the exercise content of this system also includes upper limb control training. We expect that this combination of neuromodulation and exercise training modalities, in addition to improving the ROM of the shoulder joint, will cause improvements in cortical excitability by increasing proprioceptive inputs, which in turn will affect the patient’s pain and performance.

Strengths and limitations

This research protocol has several important strengths. Its most notable feature is the innovative digital rehabilitation system that uses VR and motion capture technology to guide exercises and measure how accurately patients move. This approach is designed to make training more engaging and personalised for patients, which may help them stay motivated and perform exercises correctly, potentially leading to better recovery. The study also uses appropriate blinding to reduce bias. Only the researcher in charge of randomisation is aware of group assignments, while all other researchers and participating patients are unaware of who is receiving which intervention. The protocol is designed in a patient-centred manner to ensure that the treatment is consistent and reproducible. Furthermore, a range of assessment methods is used to thoroughly evaluate the effects of the intervention from multiple perspectives.

Nonetheless, several limitations of this protocol should be acknowledged. First, although a self-administered exercise adherence questionnaire was used, its results are susceptible to subjective influences such as recall bias or overreporting.⁴² Therefore, questionnaire data will serve only as a supplementary reference. Should any discrepancy be identified between self-reported data and actual practice—through clinical interviews or activity records—the questionnaire results will be excluded from analysis to ensure accuracy. Second, despite efforts to optimise the motion capture system, certain technical challenges remain inherent to current technologies, such as sensitivity to complex environmental backgrounds. To mitigate this, all assessments will be conducted in a controlled environment with consistent lighting and minimal clutter. Moreover, movement accuracy scores will be averaged over 15 training sessions to reduce random errors. Finally, although the calculated sample size is sufficient for detecting overall group differences, it may still be underpowered for subgroup analyses or investigating rare outcomes. As this study serves as a preliminary evaluation of the system’s efficacy, future large-scale trials with extended follow-up periods will be conducted if positive outcomes are observed at the 3-month time point—a clinically relevant time frame for deciding whether to continue or adjust treatment strategies.¹⁹