Exercise into pain in chronic rotator cuff related shoulder pain: a prospective single-group feasibility study

Abstract

Objectives

This study evaluated the feasibility of exercising into pain in rotator cuff related shoulder pain (RCRSP), data collection procedures, feedback from physiotherapists and patients, and clinically important changes in patient-reported outcome measures (PROMs).

Design

Unblinded non-randomised single-group study.

Setting

Physiotherapy clinic in Belgium.

Participants

Twelve patients with unilateral RCRSP for minimum 3 months, aged 18–65 years.

Interventions

Twelve weeks of four individualised exercises, with nine physiotherapist-led sessions with pain ratings 4–7 out of 10 on a verbal Numeric Pain Rating Scale for 9 weeks and then pain ratings 0–2 for 3 weeks. Every physiotherapy session included 15 min of manual therapy. Non-supervised exercises were: 2×/week in weeks with physiotherapy session, 3×/week in weeks without physiotherapy session.

Outcome measures

Primary: adherence, where patients were considered adherent with 78% (7/9 sessions) attendance for supervised sessions and 81% (22/27 sessions) completion for non-supervised exercises, and Shoulder Pain and Disability Index (SPADI); secondary: fear-avoidance behaviour, fear of pain, physical outcomes (strength, range of motion, scapular dyskinesis); others: ultrasound (US) imaging outcomes (acromionhumeral distance, supraspinatus tendon thickness, occupation ratio), global perceived effect (GPE). PROMs were collected via online survey, except for the GPE (via closed envelope). US measures were taken after physical measures.

Results

Adherence and adverse effects were analysed in patients who had the possibility to attend minimum seven supervised sessions (n=8): 88% of them adhered to supervised sessions, 50% to non-supervised exercises; none of them withdrew from the study, three of them obtained individual clinically important improvements in SPADI score above 20 points. The measurement protocol of physical and ultrasonographic outcomes took around 60 min.

Conclusions

Adherence to supervised sessions was satisfactory, the adherence to non-supervised exercises must be improved. Data collection procedures were feasible to perform, but some changes are recommended.

Trial registration number

NCT04154345.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Satisfactory adherence to non-supervised exercises if patients completed 22/27 (81%) of sessions.

• Satisfactory adherence to supervised sessions with attendance of 7/9 (78%) physiotherapy sessions.

• An intensive collection of patient-reported outcome measures on pain, function, fear-related behaviours and objective measures.

• The intake of analgesic drugs was not registered, but it might be relevant as non-planned cointervention.

• Absence of blinding (no control for detection or performance biases) or randomisation (no assessment of willingness to randomisation or group allocation).

Introduction

Shoulder pain is the third most frequent musculoskeletal complaint,¹ with a yearly prevalence ranging between 5% and 47%.² Physiotherapy accounted for 60% of the mean healthcare costs in a Swedish cost-of-illness study.³ Rotator cuff related shoulder pain (RCRSP) is the most reported shoulder disorder encompassing impingement, subacromial pain, rotator cuff tendinopathy, tendinosis, tendinitis and partial or non-traumatic full-thickness rotator cuff tears.⁴ Different structures might be involved in the aetiology of RCRSP, such as subacromial bursa, acromion, rotator cuff tendons and muscles,⁵ as well as different mechanisms, going from tendon overload to central sensitisation,⁶ and therefore, various types of treatment have been proposed.⁷ A non-operative intervention is the first option for the management of RCRSP, with a strong recommendation for exercise.⁷ Loaded exercises are considered safe⁸ and higher repetitions and sets seem to have superior effect compared with low dosage.⁹ However, questions remain regarding what is the best exercise modality and approach and whether pain should be elicited or avoided during exercise.¹⁰ Indeed, it is suggested that ‘exercising into pain’ can give small but significant benefits in the short term in chronic musculoskeletal disorders and should not be considered an obstacle to positive outcomes.¹¹

The theoretical rationale behind the concept of ‘exercise into pain’ is based on the positive impact on the central nervous system (CNS),^{11 12} in which exercise induces endogenous hypoalgesia due to a release of endogenous opioids and the activation of spinal inhibitory mechanisms.¹³ Protocols using exercise into pain usually included higher loads or levels of resistance, which eventually gave greater improvements in pain reduction following a dose-response effect.^{10 11} Therefore, painful exercise may offer greater benefits in short term because of a greater exercise-induced hypoalgesia.^{11 14} Furthermore, painful exercises may serve as painful conditioning stimulus to initiate the conditioned pain modulation (CPM) response, which activates descending pain inhibitory responses decreasing pain-related fear and the activity of the amygdala.¹⁴ Therefore, temporary reproduction of patient’s symptoms within a framework of ‘hurt not equalling harm’ might help to address fear avoidance and catastrophising beliefs in chronic musculoskeletal pain.^{11 12}

When we consider the literature relating to painful exercise, strength training for rotator cuff and scapular stabilisers which allowed pain during exercise was better than non-specific shoulder or neck exercises, with considerably more patients in the specific exercise group withdrawing from the waiting list for surgery.⁵ A pilot study using painful eccentric exercises showed significant results in nine patients awaiting surgery, with five patients choosing to not undergo surgical treatment.¹⁵ However, when heavy load eccentric training allowing pain during exercise was added to a rotator cuff strengthening programme,¹⁶ this did not result in superior reduction of pain or functional improvement. When comparing one painful self-managed exercise to usual physiotherapy, which included manual therapy, massage or other interventions, there were no significant between-group differences in clinical outcomes.¹⁷

Tissue irritability is also an essential factor in exercise therapy, and it might be especially important when prescribing exercises into pain. Although tissue irritability has been included in different clinical models or classifications systems for shoulder rehabilitation,^{4 18} it has not been specifically addressed in clinical trials. The threshold of pain that was allowed during exercise varied in previous studies. Patients should feel some pain during exercises,^15–17 but no more than 5 out of 10 on Visual Analogue Scale¹⁶ or on Numerical Rating Scale (NRS),⁵ with the pain subsiding by the next exercise session,⁵ by the next morning¹⁶ or directly after the exercises.¹⁷ Although pain was allowed or even recommended during exercises, studies usually do not indicate a minimal amount of pain in VAS or NRS, even when comparing specifically painful and non-painful treatments in RCRSP.¹⁹

There are various possible mechanisms why a resisted exercise programme into pain may induce pain reduction and bring about functional improvement, including changes in CNS processing, reconceptualisation of fear-related movement, and strengthening of deconditioned muscle tissue. Since different modalities of exercises have been found equally effective in RCRSP, we hypothesise that different exercises prescribed with high range of pain (4–7 on verbal NRS), could give better results than non/slightly painful modality (0–2 on verbal NRS scale). However, before testing this hypothesis in a randomised controlled trial, we conducted a feasibility study. Hence, the objectives of this study were as follows: (1) to assess the rate of adherence and adverse effects for patients receiving the intervention; (2) to describe data collection procedures; (3) to report feedback from both patients and physiotherapists and (4) to analyse the effect of exercise into pain on shoulder pain and disability, fear of pain or fear-avoidance beliefs.

Methods

Study design and participants

This feasibility study was reported according to the Consolidated Standards of Reporting Trials 2010 statement: extension for pilot and feasibility studies²⁰ (online supplemental additional files 1 and 2). It was designed as prospective single-group study with three time points of measurement: before (T0), during (6 weeks—T1), at the end (12 weeks—T2) of the intervention. The eligibility criteria are presented in table 1.

Table 1.

Eligibility criteria

Inclusion criteria

Exclusion criteria

Age 18–65 years Pain for at least 3 months Pain in the anterolateral shoulder region Pain at rest maximum 2 out of 10 on verbal NRS Patient had to test positive at least 3 out of 5 symptoms-provoking tests: pain during Neer test, Hawkins-Kennedy test, Jobe test, painful arc between 60° and 120°, pain or weakness during external rotation resistance test.36 All types of occupations were included: students, workers (including overhead workers or heavy duty workers), people on sick leave and retired people.

Bilateral shoulder pain Corticosteroid injections less than 6 weeks prior to the enrolment Participants who were pregnant, not able to understand Dutch Clinical signs of full-thickness rotator cuff tears (positive external and internal rotation lag tests or drop arm test) Evidence of adhesive capsulitis (50% or more than 30° loss of passive external rotation)37 Previous cervical, thoracic or shoulder surgery; recent fractures or dislocations on the painful shoulder Symptoms of cervical radiculopathy as primary complaint (tingling, radiating pain in the arm associated with neck complaints) Primary diagnosis of acromioclavicular pathology, shoulder instability Previous medical imaging confirming full-thickness rotator cuff tears or calcifications larger than 5 mm Patients with competing pathologies (inflammatory arthritis, neurological disorders, fibromyalgia, malignancy) Participants performing overhead sport activities for more than 4 hours/week

NRS, Numerical Rating Scale.

Procedure

Two physiotherapists recruited the participants in a private physiotherapy practice in Belgium between November 2019 and February 2020. Patients who expressed an interest in participating were given an appointment with the principal assessor. The first assessment session (T0) included: project explanation (informed consent, exercise diary, online survey), screening, assessment of physical (strength, range of motion, scapular dyskinesis) and US outcomes (acromiohumeral distance, supraspinatus tendon thickness, occupation ratio). The order of measurements was standardised, with US measures taken after the physical outcome measures. The physical and US measures were evaluated again at 6 (T1) and at 12 weeks (T2). The time planned was approximately 90 min for T0 and 60 min for T1 and T2. The patient-reported outcome measures (PROMs) were sent through online platform Qualtrics (Qualtrics, Provo, Utah (USA), versions 2019, 2020).²¹ They had to be completed within 2 days from the assessment day, whereafter an email reminder was sent if necessary.

Intervention

Patients followed a 12-week intervention with nine supervised physiotherapy sessions which included one supervised session per week and two non-supervised home exercise sessions per week. The first five supervised sessions were scheduled on the initial 5 weeks of treatment. The other four supervised sessions were spread over the following 7 weeks. During the unsupervised weeks, the patient had to practice home-based exercises three times per week at home.

Every physiotherapy session lasted about 30 min and included 15–20 min of exercise therapy (exercising into pain) and 10–15 min of manual therapy (focusing on stretching of the posterior soft tissues of the shoulder). The manual treatment of posterior shoulder soft tissues was based on two main reasons. First, the presence of posterior capsular tightness has been detected in RCRSP and other shoulder disorders when comparing healthy and affected shoulders.²² Second, restoring flexibility deficits might help to adjust scapular malpositioning, which is often present in RCRSP.²³ Moreover, the manual stretching of posterior shoulder tissues was embedded in the routine treatment of shoulder pain in the recruited private practice and therefore its implementation was expected by treating physiotherapists to be part of the intervention protocol. Guidelines were given to the physiotherapists concerning pain intensity, direction of movement, frequency of exercises. A list of possible exercises was proposed and discussed, which included closed kinetic chain exercises, exercises with elastic bands or weights. However, the physiotherapists were free to choose the type of exercises as long as the patient could reproduce it at home.

Each patient had four individualised exercises, reported according to the Consensus on Exercise Reporting Template²⁴ in online supplemental additional file 3. The physiotherapists targeted both rotator cuff and periscapular strength with a set of four exercises. Lifestyle advice, ergonomics advice, patient education regarding exercise in RCRSP were provided, in line with current physiotherapy practice in Belgium.²⁵

The patient performed four exercises chosen by the physiotherapist and the pain had to be between 4 and 7 on verbal Numeric Pain Rating Scale (NPRS) in every exercise. One exercise was in a specific painful direction, meaning in flexion, abduction or external/internal rotation or a combination of these directions. The other three exercises elicited pain between 4 and 7 but not in the same painful direction of movement. However, the pain might fluctuate between different exercises but the session was considered into pain when the average of the four exercises recorded by the physiotherapist was between 4 and 7 on NPRS. In the last 3 weeks of the treatment the pain during exercise ranged between 0 and 2 on NPRS to allow the patient to exercise in a less painful range after neuromuscular adaptations occurred in the previous phase. The pain range was set at the physiotherapy session, but it could decrease or increase during home exercises. The physiotherapist checked and adjusted the exercises at the correct pain range at every session.

Objectives and outcomes

This study had four main objectives to investigate: rate of adherence and adverse effects (objective 1), data collection procedures (objective 2), feedback from patients and physiotherapists (objective 3), effect on PROMs (objective 4).

Objective 1: rate of adherence and adverse effects

Every patient filled in an exercise diary, specifying the number of sets and repetitions for every exercise and the level of pain before, during and after 1 hour of each exercise session. The physiotherapists were also instructed to fill in a questionnaire covering attendance of patients, and for each patient the type of exercises, number of sets and repetitions, intensity (weight or colour of elastic band) and level of pain before, during and after 1 hour of each exercise session. The pain after 1 hour during the physiotherapy session was asked to the patient at every following session. The outcomes related to objective 1 are presented in box 1.

Box 1. Outcome measures for objectives 1, 2, 4.

Outcomes for objective 1

• Adherence to physiotherapy treatment (primary outcome): when patients attended at least 7 of 9 (78%) sessions.

• Adherence to non-supervised exercises (primary outcome): when patients completed at least 22 of 27 (81%) sessions. One session of non-supervised (home) exercise was considered completed when at least 80% of the total amount of sets and repetitions were executed as prescribed by the physiotherapist.

• Adverse effects: when patients were leaving the study because of treatment-related reasons. Increased pain after 1h hour of the exercise compared with baseline level was registered but not considered as an adverse effect in the final analysis, since the intervention was deliberately provocative.

Outcomes for objectives 2 and 4

• Patient-reported outcome measures (PROMs) (primary outcome): Shoulder Pain and Disability Index; PROMs (secondary outcomes): Fear-Avoidance Beliefs Questionnaire (with subscale of physical activity FABQ-PA and subscale of work FABQ-W), Fear of Pain Questionnaire-9 items; other PROMs: global perceived effect on recovery (GPE) and GPE on satisfaction.

Outcomes for objective 2

• Physical outcomes (secondary outcomes): passive range of motion in external rotation, internal rotation, scaption; active range of motion in external rotation, internal rotation, scaption; strength in scaption, external rotation, internal rotation; scapular dyskinesis (at rest, unloaded and loaded, with scapular correction tests).

• US outcomes (other outcomes): acromionhumeral distance at rest, supraspinatus tendon thickness, occupation ratio.

Objective 2: data collection procedures

The time needed to collect the data was tested for screening and objective measures (physical and US measures). The clinical questionnaires were sent via online survey Qualtrics before the treatment, at 6 and 12 weeks, except for the global perceived effect (GPE), which was collected as a measure of treatment effect in a closed envelope by the physiotherapist after 1 week of treatment, at 6 and at 12 weeks. The outcomes related to objective 2 are presented in box 1 in the order of measurement. Details on PROMs and on the measurement protocol of physical and US measures are described in online supplemental additional file 4.

Objective 3: feedback from patients and physiotherapists

The patients who attended at least seven sessions out of nine were interviewed in a face-to-face meeting at 6 and 12 weeks by the first author to evaluate their experience with the intervention and the adherence to both the supervised and non-supervised sessions. If a face-to-face meeting was not possible, the patient was interviewed by phone. The exercise and pain diary were filled out by the patient and explored during the assessments. Feedback from the physiotherapists was explored at the end of the study period.

Objective 4: effect on PROMs

The PROMs are indicated in box 1 and relative details are reported in online supplemental additional file 4. Concerning the primary patient-reported outcome Shoulder Pain and Disability Index (SPADI),²⁶ the Minimal Important Change (MIC) was calculated for each patient as a change of at least 43% of the individual baseline scores, as proposed by Thoomes-de Graaf et al.²⁷ A change of less than 20 points might be due to measurement error.²⁷

Sample size

For the primary objective, a sample size of 12 patients was calculated to test 80% of compliance rate, ranging between 0.78 and 0.84 within a 95% of CI, obtained with the ‘score method incorporating continuity correction’ reported by Newcombe et al.²⁸ We considered in this calculation 36 sessions (9 supervised and 27 non-supervised sessions), 12 exercises in total (4 different exercises, 3 sets for each exercise) for a total of 432 observations.

Randomisation and blinding

No control group was present, and randomisation or blinding was not implemented.

Patient and public involvement

Patient involvement was important in the intermediate (at 6 weeks) and final (at 12 weeks) analysis of the project as one of the reasons we conducted the feasibility study was to obtain patient feedback on the acceptability of the exercise protocol. Patients who could attend at least seven out of nine sessions were interviewed at 6 or 12 weeks follow-up in person or, when this was not possible, by phone. They were asked about the reasons why they could not exercise at home or why they did not attend a physiotherapy session. They were not involved in results dissemination.

Data analysis

Demographics, patient characteristics and recruitment time were analysed with Microsoft Excel (2016) and JMP Software (V.15.2.1, SAS Institute). Feedback from patients and physiotherapists was noted down and reported in short interviews, but no formal registration or transcription were conducted. The PROMs, physical and US outcomes were reported quantitatively. Change in the primary patient-reported outcome (SPADI) was described in relation to the individual MIC: patients with change scores of at least 43% of their baseline SPADI were considered clinically improved.²⁷

The physical and US outcomes were included for evaluating data collection procedures and only pretreatment data are presented. The continuous variables were described as median, minimum or maximum values or IQRs due to the small sample size.

Results

Participant flow and recruitment

Two physiotherapists screened 65 patients with RCRSP, excluding 49 patients in a first assessment (figure 1), while the principal assessor excluded 4 extra patients in the final assessment for eligibility. However, the follow-up measurements and the physiotherapy sessions during the pandemic of COVID-19 were stopped in March and April 2020, and therefore, physiotherapy treatments and measurements of physical and US outcomes were lost for seven patients at 6 or 12 weeks. In the analysis of adherence and adverse effects, we considered only the patients who had the possibility to attend a minimum of seven supervised sessions (n=8). They continued to exercise at home also during the period of lockdown and two patients completed their sessions during the lockdown and they were monitored by the physiotherapist by phone. The data on PROMs were also analysed (n=8). The time and order of data collection for US and physical outcomes were analysed for all patients (n=12), but only pretreatment data are reported as part of the feasibility of the measurement protocol. Feedback from physiotherapists (n=2) and patients analysed (n=8) was elaborated in face-to-face meetings or by phone.

Figure 1.

Flow chart adapted from CONSORT flow chart. CONSORT, Consolidated Standards of Reporting Trials.

Twelve subjects participated in the study and their baseline characteristics are reported in table 2. Although only 2 participants had heavy workload, identified as full-time occupation demanding frequent overhead activities, 8 of them had some pain during work, while 10 had pain during sport and/or leisure activities. The PROMs, physical and US outcomes for 12 patients at baseline are reported in online supplemental additional file 5. The GPE of recovery and satisfaction was collected separately in a closed envelope after 1 week of treatment for the first time. Data at this time point are presented in online supplemental additional file 5.

Table 2.

Baseline characteristics

Variable	n=12
Gender, female	7 (58.3%)
Age, years	50.5 (16.5)
Body mass index, kg/m2	23.2 (3.4)
Working status
Working	11 (91.7%)
Student	1 (8.3%)
Duration of symptoms, months	6.5 (11.3)
Dominant side affected	7 (58.3%)
Previous treatments	8 (66.7%)

Data are presented as median (IQR) or n (%). ‘Previous treatments’ included any treatment for shoulder pain (such as physiotherapy, injections more than 6 weeks ago, osteopathy) conducted before the enrolment.

Objective 1: rate of adherence and adverse effects

Eighty-eight per cent (7/8) of patients fulfilled the criteria of attending at least 7/9 of the supervised sessions. Only 50% (4/8) completed at least 22/27 of the non-supervised exercises, as prescribed by the physiotherapists in terms of repetitions and sets. One patient did not attend three times the physiotherapy sessions because of work-related reasons, while five patients could not exercise at home as prescribed, because of sickness (n=1), increased pain related to return to sport (n=2), lack of time (n=1) and misunderstanding with the physiotherapist concerning the number of repetitions and sets (n=2). None of the patients considered in this analysis withdrew from the study.

Considering the mean of pain in four exercises for every supervised session in the first 9 weeks, four patients (57%) trained between 4 and 7 on verbal NRS, while three (43%) did not reach this range. The questionnaire for one patient was not completely filled in by the physiotherapist, and therefore, not considered in this analysis. The number of sessions of these patients varied between 4 and 8 sessions in the first 9 weeks, depending on the availability of the patients.

Objective 2: data collection procedures

Enrolment and screening for eligibility criteria took around 30 min while the measurements for all the physical and US measures took 60 min. Filling out the online questionnaires took on average 6 min at the first time, 7 min the second time and 4 min the last time. Reminders were sent out a maximum of two times per person. The online survey was more effective than closed envelope as modality of data collection, since 5 data were lost at baseline with the envelope modality for GPE (online supplemental additional file 5). Patients forgot to give the envelope back to the physiotherapists, especially during period of restrictions due COVID-19 pandemic. Concerning the loaded scapular tests, the weight of 1 kg was very provocative in two patients. Since the movement was repeated five times in abduction and five times in flexion, these patients performed the tests with 0.5 kg instead.

Objective 3: feedback from patients and physiotherapists

Physiotherapists had difficulties in providing four painful exercises for 9 weeks for different reasons. First, there was an increase in pain during the four exercises, which was becoming unbearable for some patients at the fourth exercise, especially in the first sessions. Consequently, there was a motivational issue for these patients who had high initial pain levels. Other patients had fast recovery and it was not possible to provoke pain adding more loads in four different exercises already at the third or fourth week. Furthermore, some issues during home exercises which could have influenced the pain perception occurred. For example, some patients trained one extra session during the week, or they did not wait 24 hours between sessions. Two patients increased their sport activities during the treatment period, influencing the pain during home exercises. Moreover, one patient had an extra physiotherapy session during the study period and another patient performed additional stretching exercises. Although the return to sport could influence the level of pain during exercise, some patients were willing to engage into sport while some others were too afraid. Therefore, the physiotherapists did not give limitations on this matter but they let the patients decide based on their willingness.

Objective 4: effect on PROMs

Of those patients who had the possibility to attend at least seven out of nine supervised sessions (n=8) before the pandemic of COVID-19, results of PROMs are shown in online supplemental additional file 5 and figure 2. Considering the SPADI change for each individual patient, three patients had a significant individual change which was also superior to the measurement error of 20 points, while one patient had a significant change but it was inferior to 20 points (see online supplemental additional file 5). In figure 2, improvement in pain and function is seen in all patients, as indicated by decreased SPADI score, while other PROMs showed different trends over time depending on the patient.

Figure 2.

SPADI, FPQ-9, FABQ-PA, FABQ-W at 0, 6 and 12 weeks follow-up. Data of SPADI, FPQ-9, FABQ-PA, FABQ-W for eight patients are converted to a value from 0 to 100, where 100 identifies a worse score. FABQ-PA, Fear-Avoidance Beliefs Questionnaire-Physical Activity; FABQ-W, FABQ-Work; FPQ-9, Fear of Pain Questionnaire; P1–P8, ID Patient; SPADI, Shoulder Pain and Disability Index.

Discussion

This feasibility study showed that a significant proportion (43%) of patients did not adhere to a programme of exercise into pain of 4–7/10 for nine consecutive weeks. The rate of attendance of physiotherapy was satisfactory, but the rate of adherence to non-supervised exercises was not. The time for data collection complied with the prespecified time frame, and the delivery of the clinical questionnaires via online survey was both practical and achievable. The order of physical outcome measures and modalities of scapular testing should be adjusted in future studies.

According to the feedback of the physiotherapists, it was not possible to prescribe four painful exercises for 9 weeks for all patients. Physiotherapists encountered problems in motivating the patients to exercise into pain or could not find painful exercises for some patients. This suggests that ‘exercise into pain’ might not be applicable to all patients. This was not the case in a similar study conducted by Vallés-Carrascosa et al,¹⁹ as they did not report any drop-out in both painful and non-painful exercise groups. However, patients in the painful group trained below 40 mm on VAS scale carrying out only one painful eccentric exercise of the supraspinatus out of six exercises. Between-group differences in pain or function were not found at the end of the treatment. One might argue that the difference in VAS during training between groups was not sufficient to elicit significant differences, or maybe that different pain levels were not a relevant factor in the study. Since it was not possible to maintain four painful exercises for 9 weeks in the current study, we propose that exercise into pain should be limited to one exercise in the painful direction with clear limits during training (4–7 on NRS), while the remaining three exercises could be performed at a pain level between 0 and 2 on NRS. On the other hand, it is possible that this exercise protocol could work only in a subgroup of patients continuously supervised by the physiotherapist and who are very motivated to exercise, even into pain, as self-efficacy has shown to be an important predictive factor during therapy.²⁹ Moreover, a qualitative analysis is suggested for future studies to understand the barriers and beliefs around the concept of ‘exercise into pain’ for both patients and physiotherapists.

Second, adherence to home exercises was lower than expected. Closer monitoring and discussion of exercises and expectations between physiotherapists and patients during future studies is recommended. Images and videos of exercises might enhance adherence and help the patient in the performance of the home exercises. To deal with the low rates of patient adherence, telerehabilitation may be a good approach. It has shown good results (92%) combined with usual care when applied in people with RCRSP, compared with usual care alone (67%).³⁰ Moreover, the intake of analgesic medication during treatment should be registered in future feasibility or pilot studies as the flare-ups of pain inducing a patient to take analgesic drugs after training session might be considered as a non-planned cointervention and a medication use may be a significant variable that could impact on results.

Considering the individual SPADI changes, only three patients had a significant MIC which was also above the measurement error of 20 points. However, the median change in SPADI score for all patients was 29 points, which was higher than previous estimated values (8–13.2 points)³¹ and also higher than another recent feasibility study with similar RCRSP population (17 points) by Major et al.³² Although patients were similar in terms of age (around 50 years old) and gender distribution (higher prevalence of women), other differences may have influenced the results, such as previous (failed) treatments in the study by Major et al.³² Moreover, our population presented higher SPADI at baseline compared with Major et al, which could have led to greater changes during time.

The choice of the research team to not use a specific set of exercises but rather to use a list of possible exercises was motivated by two main reasons: to allow the physiotherapist to prescribe individualised exercises and to adapt the exercises in the provocative range. We believe that this approach to chronic shoulder pain is more applicable to routine clinical practice rather than only one specific modality of exercise, which will not fit all patients. This feasibility study suggests that challenging patients with chronic RCRSP into pain-provoking exercises in frequent supervised physiotherapy sessions may bring significant individual results in pain and function (SPADI score) in some patients, but not in all of them. After all, this type of exercises might work only on a subgroup of patients who are highly motivated and we cannot estimate causation in a single-arm non-randomised study. We have to be very cautious as these effects could be obtained by the natural course of the intervention and/or by the placebo effect as we did not specifically analyse contextual (ie, patient–therapist relationship) or non-specific (ie, natural course) effects.³³ Moreover, low adherence to non-supervised exercise indicates that not all patients wanted or could exercise into pain. As we included patients with chronic symptoms and low tissue irritability at rest, our results could be generalised only to patients with similar characteristics. Adherence to painful exercises could have been lower in patients with acute symptoms and/or highly irritable tissues, but we could only speculate on that as we did not test it.

A standardised order of US measurements at the beginning of the testing is suggested for future studies. The reason is that some physical outcome measures, in particular the measurements of maximal strength, were provocative and therefore 1 hour of testing could affect both supraspinatus tendon thickness and acromio-humeral distance on the US image. A significant increase in supraspinatus tendon thickness has been previously detected after 1 hour of fatigue loading exercises in patients with rotator cuff tendinopathy, although it did not reach the minimal detectable change.³⁴ The acromio-humeral distance also reduced significantly 1 hour after the fatigue loading in the same study. Since both these US parameters can be significantly affected after 1 hour of exercise, it is suggested that US measurements should be carried out at the beginning of the testing procedure. Regarding loaded scapular tests, the weight used for the testing was reduced for two patients because of high pain reported during repetitive testing. Therefore, it is suggested to set the provocative load at the first assessment at 0.5 kg for all patients and gradually increase the load.

Limitations

This feasibility study had some limitations. First, we aimed to recruit patients with potentially low tissue irritability, excluding participants with resting pain higher than 2 out of 10 on verbal NRS. However, tissue irritability can be rated in more detail (ie, absence of night pain, minimal pain with overpressure) as part of the Staged Approach for Rehabilitation Classification: Shoulder Disorders (STAR-Shoulder).¹⁸ This system of rating tissue irritability in low, medium and high levels recently showed acceptable reliability and strong relationship with patient-reported outcomes,³⁵ and it is suggested for future studies. Based on the level of tissue irritability, this approach can help the clinicians in the choice of training intensity and treatment strategy.¹⁸ Second, the US outcomes were measured after the other physical tests, but this can be a limitation as the measures of acromio-humeral distance and supraspinatus tendon thickness might be affected after 1 hour of fatigue loading exercises.³⁴ Although the measurement protocol was not addressing specifically fatigue loading exercises, it is suggested to measure first the US outcomes and then the other physical tests to avoid that measures such as repetitive strength testing influence US measures. Moreover, feedback from patients was not registered or transcribed and the interview was performed by a researcher who was involved in the inclusion and assessment of patients. A more in-depth and well-structured interview performed by a person not involved directly in the research could give better insight on the qualitative value of this type of study. Lastly, the sample size of this study was small and the number of lost measures was high, but this was mainly due to unforeseen consequences of the COVID-19 pandemic and related restrictions.

Conclusion

This feasibility study showed that not all patients were adherent to exercise into pain for 9 weeks and that the adherence to non-supervised exercises should be improved. However, there was an acceptable rate of attendance to the physiotherapy sessions and no patients withdrew from the study. Data collection procedures were feasible and achievable, but the order of measurement in the protocol and modalities of scapular testing should be adjusted in future trials.

Data availability statement

Data are available on reasonable request. The datasets used during the current study are available from the corresponding author on reasonable request.

Ethics statements

Patient consent for publication

Consent obtained directly from patient(s).

Ethics approval

This study involves human participants and was approved by the Ethics Committee of the Antwerp University Hospital approved this study (ref: B300201837376). Participants gave informed consent to participate in the study before taking part.