The Strengthening Exercises in Shoulder Impingement trial (The SExSI-trial) investigating the effectiveness of a simple add-on shoulder strengthening exercise programme in patients with long-lasting subacromial impingement syndrome: Study protocol for a pragmatic, assessor blinded, parallel-group, randomised, controlled trial

Resistance training, aimed at strengthening the rotator cuff muscles and scapula stabilising muscles, is an important component of most novel exercise interventions for long-lasting SIS [9, 12, 13]. This seems relevant as patients with SIS have significant force impairments in both the glenohumeral and the scapulothoracic joint [4, 14, 15]. However, these force impairments are most pronounced in the glenohumeral joint, with a 33% deficit in external rotation force and a 29% deficit in abduction force compared to only 8% and 18% force deficit in protraction and retraction of the scapula, respectively [15]. This suggests that specific training of the glenohumeral muscles is especially relevant. Such specific resistance training of the, often degenerated [16], rotator cuff muscles and tendons also seems relevant as this training modality is known to improve muscle and tendon health through various pathways [17].

It is uncertain if the novel exercise intervention programmes, aimed at patients with long-lasting SIS, are focusing sufficiently on strengthening exercises. Accordingly, changes in muscle force are only tested in a few randomised controlled trials (RCTs), including resistance training in the rehabilitation of patients with SIS [12, 13, 18, 19]. In these studies, maximum force in external rotation increased by only 4–15% in average from baseline to follow-up in patients doing strengthening exercises [12, 13, 18, 19], which is far from restoring the 33% impairment in external rotation force previously reported [15]. Though strength gain may not be the primary aim of the intervention, the sparse effect on this objective outcome points towards a possible gap in the rehabilitation of patients with long-lasting SIS. This suggests that the exercise dose received by patients might have been too small, either because the prescribed resistance training programmes were too mild, or simply due to lack of adherence to the prescribed programmes. The trial described in this protocol will aim to investigate the effectiveness of adding a simple shoulder strengthening exercise programme to usual care in patients with long-lasting SIS.

To further ensure that we do not initiate a redundant trial, we have performed a systematic literature review. On January 26, 2016, we performed a systematic literature search on controlled trials investigating the effect of resistance training in patients with SIS. We searched Medline via PubMed (86 hits) and Embase via OVID (38 hits) using search strings for condition (shoulder impingement) and intervention (resistance training). We also searched www.​clinicaltrials.​gov for registered (finalised, ongoing or planned) trials. We only identified one trial [18] in which the effect of resistance training was isolated as the active part of the intervention. Additionally, in a recent systematic review, only one study [20] investigating the effect of an exercise intervention in patients with persistent SIS was identified, and in that study the exercise intervention was compared to surgery. In conclusion, the existing evidence regarding the effect of resistance training in patients with persistent SIS is very limited and we therefore consider this to provide a clear-cut ethical, scientific and economic justification for the trial described in this protocol.

As previously stated, the exercise dose received by patients might have been too small in previous studies, either because the prescribed resistance training programmes were too mild or simply due to lack of adherence to the prescribed programmes. In the study by Lombardi et al. [18], patients allocated to an intervention consisting of supervised strengthening exercises only increased 9% in external rotation strength (derived from Lombardi et al. [18]), which was not significantly different from the waiting list control group. As exercise sessions were supervised, this absence of effect is likely a consequence of the prescribed dose being too small, with exercises for flexion, extension, internal and external rotation performed twice a week in 8 weeks with 2 × 8 repetitions at 50% and 70% of 6RM, respectively. One way to efficiently increase the exercise dose in rehabilitation is to include home-based exercises, as was done in a study by Bennell et al. [12]. In that study, an intervention consisting of manual therapy and home exercises was found superior to placebo in improving abduction strength (significant 15% increase), but not external rotation strength (non-significant 4% increase), pain scores or patient-reported function at the end of treatment. The limited gains in shoulder strength observed in the study by Bennell et al. [12] might also be a consequence of an insufficient exercise dose received by the patients. However, it is difficult to determine whether the sparse improvements in shoulder strength are due to the programme being too mild or because of low adherence, as the amount of external resistance applied to exercises is not described. Nevertheless, exclusion of non-adherent patients from the analyses did not alter the results, indicating that the sparse improvements in shoulder strength were not only due to low adherence.

In the current study, we aim to increase the received dose of strengthening exercises aimed at the rotator cuff muscles in patients with long-lasting SIS without aggravating their symptoms and with minimal risk of adverse events. To achieve this, the intervention will start with exercises using low relative resistance but high volume (number of sets per session and frequency of sessions). This approach is also relevant, as resistance training with low intensity/resistance, high frequency and approximately 3–4 sets per muscle group is associated with the highest gains in muscle strength in untrained individuals [21]. All sets are to be continued to failure, as this maximises the gain in local muscle endurance [22]. Furthermore, the low resistance will minimise pain during exercises and protect the muscle from overload injuries, ensuring that all patients are able to complete the programme.

The focus of exercises is on isometric holds and slow dynamic strengthening with a long time under tension (TUT). A high TUT is achieved through exercises with a low contraction velocity and the inclusion of an isometric component to increase the exercise stimulus without increasing external resistance; this in turn reduces the peak load on the involved, and possibly damaged [16], muscles and tendons. Thus, exercises become safer and less painful, which again will allow patients to actually adhere to the exercise intervention. Furthermore, the isometric part will be performed with the shoulder in approximately 30–45 degrees of scaption, the position in which the compression forces on the supraspinatus muscle and tendon are lowest [23, 24]. Exercises will mainly focus on external rotation and abduction, functions in which both supraspinatus and infraspinatus muscles are highly active [25].

Patients with long-lasting SIS have lower mechanical pressure pain thresholds (PPT), both at the shoulder region and in other anatomical regions (such as the tibialis anterior), which may reflect an increased manifestation of peripheral and central pain sensitisation [26]. Pain sensitisation modulates the patients’ pain experience and, furthermore, the presence of hyperalgesia or referred pain before decompression surgery significantly worsens the outcome after 3 months [26], while more pain catastrophizing is related to persistence of symptoms in patients with long-lasting SIS [27]. It therefore seems likely that manifestations of peripheral and central pain sensitisation, and peripheral adaptations to long-lasting pain, could also affect the outcome of an exercise intervention. This could be either through an alteration in adherence to the intervention due to pain sensitisation, or through other mechanisms related to central pain sensitisation and peripheral adaptations.

The effect of scapula function on treatment outcome in patients with SIS is often debated, and one could argue that the increased emphasis on rotator cuff muscles, as is proposed in the current trial, would neglect an important aspect of the rehabilitation. Accordingly, it has been suggested that rotator cuff emphasis should only be after scapula control is achieved, as insufficient dynamic stability of the scapula could cause abnormal shoulder kinematics and impingement symptoms [28]. Assuming this, the effect of the add-on intervention in the current study, focusing mainly on strengthening exercises for the rotator cuff, could be modified be the presence of scapula dyskinesis. However, Mulligan et al. [29] recently found that treatment outcome was not significantly different between patients randomised to scapula stabilisation first, or rotator cuff strengthening first, questioning the importance of maintaining a scapula focus in the beginning of rehabilitation.

This study aims to investigate the effectiveness of adding a simple shoulder strengthening exercise programme to usual care in patients with long-lasting SIS. Therefore, the relevant control condition is usual care, which is to be compared to the intervention condition involving an add-on exercise programme and usual care.

This is relevant because usual care for patients that are covered by the Danish National Guidelines for Treatment of SIS [10], and hence have a medically justified need for general rehabilitation, is referral to general rehabilitation under the Danish health act § 140. Such rehabilitation is likely based on the most novel evidence from previously mentioned RCTs [12, 13, 18], where only limited gains in muscle strength are found. It therefore seems unlikely that the current treatment addresses the issue of glenohumeral muscle strength to a sufficient degree.

Furthermore, direct monitoring of adherence to the add-on intervention makes it possible to conduct secondary analysis, investigating the relationship between exercise adherence and changes in both strength and patient-reported function.

The SExSI-Trial is a pragmatic, assessor blinded, randomised, controlled superiority trial, with a two-group parallel design. Patients will be randomised to either usual care or a home-based intervention consisting of progressive high volume resistance training in addition to usual care with a 1:1 allocation. The primary end-point will be change in patient-reported shoulder function 16 weeks after baseline.

This clinical trial protocol is based on the PREPARE Trial guide [31] and the SPIRIT checklist [32] (Additional file 1). The trial report will adhere to the CONSORT guidelines for reporting parallel group randomised trials, the CONSORT extension for pragmatic trials http://​www.​consort-statement.​org, and the TIDieR template for intervention description and replication [33].

Participants will be recruited from the secondary care orthopaedic outpatient clinic at Hvidovre Hospital. Consecutive sampling will be used to ensure generalisability of the results. Accordingly, all patients who are referred for the first time to the clinic for examination of their current shoulder disorder lasting at least 3 months, who are living in the Capitol Region of Copenhagen, Denmark, who are not pregnant, do not permanently use strong pain medication (defined as morphine analgesic or similar), are aged 18–65 and are considered able to understand spoken and written Danish will be evaluated for eligibility by an orthopaedic specialist. All eligible patients will be provided with written information about the study.

Patients will be invited to participate if they meet the following inclusion criteria: (1) At least three positive of the five diagnostic tests for subacromial impingement syndrome (Hawkins-Kennedy test, Neer’s test, pain-full arc, Resisted External Rotation test and Jobe’s test) [34]; (2) Have been provided with or offered a rehabilitation plan due to a medically justified need for general rehabilitation after discharge from the hospital under the Danish Health Act § 140; and (3) hand in a completed SPADI questionnaire on the day of the medical examination.

Patients will be excluded if they fulfil any of the exclusion criteria, namely (1) have a radiologically verified new or previous fracture related to the shoulder joint, including the scapula; (2) have radiologically verified glenohumeral osteoarthritis; (3) have a clinically suspected luxation or sub-luxation of the glenohumeral, acromioclavicular or sternoclavicular joint; or (4) have a clinically suspected labral lesion, complete traumatic tear of the rotator cuff, frozen shoulder or other competing diagnoses (i.e. rheumatoid arthritis, cancer, neurological disorders, fibromyalgia, psychiatric illness).

Both groups will receive usual care consisting of an offer of referral to general rehabilitation in the municipal clinic under the Danish health Act § 140, sometimes with the option to instead choose a private physiotherapy clinic, partly at their own expense. The referral is standard procedure when a patient with SIS is considered to have a medically justified need for general rehabilitation, as identified during the examination at the orthopaedic outpatient department (secondary care unit). This procedure is based on the Danish National clinical guideline on diagnostics and treatment of patients with selected shoulder disorders [10], as exercise therapy is recommended as first line of treatment for SIS.

In the context of this study, usual care includes all treatment received by a patient during the time between baseline and follow-up, except that included in ‘Strengthen your Shoulder’. Therefore, usual care might include a range of treatment modalities including advice, stretching, exercises, manual therapy, massage, acupuncture and electrotherapy at the discretion of the treating physiotherapist and doctor. A description of usual care following the TIDieR guidelines [33] is included in Additional file 2.

Patients in the intervention group will, in addition to usual care, receive instructions regarding a home-based, progressive, high volume resistance-training programme immediately after randomisation. Instructions in the exercises are provided at baseline (week 0) and after 2, 5, 10 and 16 weeks. The exercise programme has been developed in cooperation with a team of experts with extensive knowledge of both exercise physiology and rehabilitation and training of patients with shoulder disorders (MC, TB, KT and MR). The design of the exercises, and their progression, is based on the relevant literature regarding exercise physiology, shoulder biomechanics and strength training principles for both healthy people and patients with shoulder disorders. A standard protocol for individual adaptation of the programme based on pain response is part of the intervention.

A thorough description of the intervention, and modifications of the intervention, following the TIDieR guidelines [33], is included in Additional file 3, and the full set of strength training descriptors, as suggested by Toigo and Boutellier [35], are presented in Table 1. An online instruction video of the exercises will be made available upon publication of the main results of the trial, and a link to these is provided in Additional file 3.

The sample size estimation is based on the primary outcome SPADI. Studies on patients with SIS have shown standard deviations (SD) for SPADI change score after 11–12 weeks of between 17 [12] and 18.5 [56], while longer durations seem to entail larger SDs, with a SD of 22 reported for SPADI change from 0 to 22 weeks [12]. The latter is similar to our own unpublished data from SIS patients, showing a SD of 22.5 for SPADI change from 0 to 6 months. Based on this previous data, we expect a common SD of 19.5 for SPADI change values from week 0 to 16, the primary outcome. For the purpose of this study, the MCID for SPADI will be considered to be 10 points, based on previous studies [45, 48]. The negative effect that any dropouts will have on the statistical power will, to some degree, be reduced by the use of multiple imputation. However, the imputation of data will not fully redeem this, as multiple imputation can cause an underestimation of the effect and a larger variation in outcomes. Therefore, we aim at having a high power of 95% to verify an effect equal to or higher than the MCID of 10 points on SPADI, at a 5% significance level. To obtain this, a total of 200 patients will be required (100 in each group). This corresponds to a power of 89.7% in case of a 20% dropout and a power of 85.4% in case of a 40% dropout.

All participants will be recruited consecutively from the arthroscopic centre at the orthopaedic department, Hvidovre Hospital immediately after undergoing clinical examination performed by an orthopaedic specialist, who will also make the initial eligibility screening and provide written information about the trial for eligible patients. In order to achieve adequate enrolment, the orthopaedic specialists, who are at the first line of recruitment, will be informed about the progress of the trial at regular formal and informal meetings.

At baseline, after written consent is signed and before randomisation, all baseline assessments (Fig. 1) will be conducted by clinical physiotherapists who are trained in all assessment procedures and mutually aligned in order to improve the quality of data. All post-allocation assessments (week 5, 10 and 16) will be performed by the same group of testers in order to secure consensus regarding the instructions in questionnaires and conduction of tests. Participants will be instructed not to take any pain medication during the 8 hours prior to any assessment. Outcome assessors will also be trained in counselling for adherence to follow-up testing (to facilitate retention) and avoidance of disclosing allocation in a uniform manner. The study instruments are all described in detail in Additional file 7.

In general, once a participant is enrolled, every reasonable effort will be made to collect all outcomes for that participant, regardless of any deviations from the intervention protocol. Participants will receive text-message reminders for all scheduled follow-up appointments. The weekly text-messages with standardised questions regarding exercise time will also ensure that participants keep in mind their participation in the study. If participants do not attend their scheduled follow-up assessment, they will be contacted and offered to re-schedule to another date. If a participant should withdraw consent to follow-up assessments, the participant is offered the option to continue with the assessment of PROs.

Reasons for non-adherence to add-on intervention will be recorded by the investigator providing the intervention, either at the intervention visits or by phone if the participant does not come to the scheduled appointment. Reasons for non-retention will be recorded by the outcome assessor if this participant withdraws during a follow-up session, or by the primary investigator in a telephone interview.

Participants will be randomly assigned to either the control or intervention group (CG or IG) at a 1:1 allocation ratio, using a computer generated randomisation schedule of permuted blocks of random sizes ranging from 4 to 10.

Participants will be randomised using sequentially numbered, opaque, sealed envelopes. Investigators taking part in allocation and data collection will be blinded to block sizes and randomisation sequence at all times during the study period. Allocation concealment will be ensured, as the envelopes will not be opened before the participant has been irreversibly included in the study.

The creation of the randomisation schedule of permuted blocks of random sizes, and subsequent packaging of sequentially numbered, opaque, sealed envelopes will be performed by persons not else involved in the trial.

The final enrolment and subsequent allocation of participants will be conducted by investigators not taking part in any outcome assessment, who will be blinded to the randomisation sequence at all times during the intervention period. Outcome assessors will not take part in any of the processes related to allocation.

Outcome assessors performing the outcome assessment at baseline and follow-up weeks 5, 10 and 16 will be blinded to group allocation. Given the nature of the intervention, which requires the treating therapist to know the intervention, blinding of intervention providers is not deemed feasible, and therefore will not be performed. In order to obtain valid results from this trial, intervention receivers will be kept blind to treatment allocation. This will be attained through minimal information about the content of the add-on intervention and control condition until group allocation is final. Accordingly, participants will be informed (written and orally) that the study compares two different treatment regimens, and that both include the treatment elements normally offered, and adhere to the clinical guideline. Additionally, study participant will be strongly inculcated not to enclose or discuss treatment allocation with the outcome assessors, medical doctors or physiotherapists providing usual care.

All pre-defined analyses will be performed by a data analyst blinded to allocation.

For the primary end-point, namely SPADI score after 16 weeks, and for all continuous secondary outcomes listed in Table 2, a constrained linear mixed model (cLMM) will be applied in order to compare the change from baseline to 16 weeks in the IG to that in the CG. The model will contain the outcome at 16 weeks as the dependent variable, treatment group (IG or CG) as the main effect and both baseline score and any additional follow-up measurements as repeated measurements, to estimate the differences in mean change between groups and corresponding 95% confidence intervals (95% CI) (Table 3). The covariance structures will be selected based on the MAICE procedure [57].

Mean scores and corresponding 95% CIs will be reported for all outcome time-points (t₀, t₁, t₂ and t₃) (Table 3). Within group changes between all outcomes assessment time-points (t₀-t₁, t₁-t₂ and t₂-t₃), and between baseline and last assessment (t₀ and t₃), and corresponding 95% CI, will be reported (Table 4). Finally, differences in within group changes between all outcomes assessment time-points (t₀-t₁, t₁-t₂ and t₂-t₃) and the corresponding 95% CIs will be reported (Table 5). Results from the repeated measures analysis with SPADI score as outcome will be visualised as illustrated in Fig. 2.

For comparison of binary outcomes, proportions will be compared using a χ² test, and odds ratio estimates and corresponding 95% CIs will be reported.

Results from analyses comparing IG and CG for all primary and secondary outcomes will be reported in the second paper outlined in the ‘Dissemination policy’ section, except for the results on secondary outcomes regarding pain sensitisation and scapula dyskinesia, which will be reported in the third and fourth paper outlined in the ‘Dissemination policy’ section, respectively.

Adverse events will be defined in this context as any unintended, unfavourable findings, symptom or illnesses that occur during the assessment or the add-on intervention, whether it can be attributed to the assessment or not. Adverse events will be recorded in part by the patient as a spontaneous recording during assessment and by open questioning.

Acute exacerbations of shoulder symptoms will be recorded by the primary investigator (MC) and, as a safety precaution, in case a medical evaluation is required, the participant will be referred to the Medical Advisor (PH). Patients in the intervention group experiencing exacerbations lasting more than 1 week will be referred to the Medical Advisor (PH) and evaluated if further medical care is needed.

Serious unexpected side effects or serious adverse events will be reported to the Capital Regional Ethics Committee in Denmark within 7 days after sponsor or the primary investigator has become aware of the incident. Serious adverse events will be categorised according to the definitions established by the United States Food and Drug Administration [59] and will be assessed by the primary investigator for possible relations with the assessment and/or intervention to consider whether there is a reasonable possibility that the adverse event can be caused by either.

No audits are planned.

All results from the study, be they positive, negative or inconclusive, will be published in international scientific journals. The project leader will enforce publication. Furthermore, the results will be presented at national and international conferences. Working titles for scientific publications are listed below. In the primary trial report, all collected outcomes will be defined and referenced to the dissemination plan below if data or analyses are not reported in the primary trial report.

Additionally, the results of this study will be communicated directly to the participants, who will be encouraged to comment on disseminated trial results in order to improve their further public dissemination. Furthermore, trial results will be disseminated to the public in general through the daily press.

Aside from the above, this trial has some methodological strengths and limitations which must be taken into account when interpreting the study findings.

Issue date: 10 May 2017, Protocol amendment 01, Author MC.