Key messages regarding feasibility

  1. 1.

    What uncertainties existed regarding the feasibility?

    • We wanted to investigate the recruitment rate, follow-up rate, objective physical activity assessment, adherence to the self-managed exercises, fidelity and adverse events of the Ad-Shoulder intervention.

  2. 2.

    What are the key findings?

    • Objective physical activity measures, follow-up, the physiotherapists’ fidelity to the intervention and patients reporting adverse events (including mild) were judged to be feasible.

    • The feasibility of recruitment and adherence to the self-managed exercise programme were below the expected level.

  3. 3.

    What are the implications of the findings for the design of the main study?

    • Further intervention development is necessary to understand whether adherence to the self-managed exercise could be enhanced.

    • Include additional recruitment sites to optimise recruitment.

Background

Shoulder pain is a prevalent and often long-lasting musculoskeletal disorder [1, 2]. Shoulder pain disorders are common in clinical practice, with an incidence rate of about 10 per 1000 in primary care [3, 4]. The impact on people with shoulder pain disorders is multi-dimensional including pain, activity limitations, social restrictions, sleep disruption, cognitive dysfunction, emotional distress and other pathophysiological manifestations [5]. Subacromial pain is the most common shoulder diagnosis, accounting for up to 70% of the cases [3, 6, 7].

Exercise is recommended as one treatment for patients with subacromial pain [8,9,10,11]. Although research suggests that exercise including some level of resistance and maintained for at least 12 weeks might be important prescription parameters, there remains a lack of knowledge about the optimal type- and dosage of exercises [12]. There is evidence to suggest that self-managed home exercises confer similar outcomes to supervised exercise programmes [12, 13]. In patients with chronic musculoskeletal pain, it has been suggested that there is an association between exercise adherence and improvements in clinical outcomes [14,15,16]. A Cochrane review on interventions to improve adherence to exercise in patients with chronic musculoskeletal pain concluded that individualised exercise therapy and self-management techniques may enhance exercise adherence [17]. There is also moderate quality evidence that behaviour change techniques, such as social support, goal setting, demonstration of behaviour, graded tasks and self-monitoring of behaviour, may improve exercise adherence among patients with persistent musculoskeletal pain [18, 19]. Despite this, behavioural frameworks to enhance adherence to home exercise programmes have been little implemented in current trials on shoulder pain [20].

In order to improve adherence to exercises in patients with subacromial pain, we developed a personalised supported self-management intervention (the Ad-Shoulder intervention). In order to assess the feasibility and acceptability of an intervention, it has been recommended to run a feasibility study, before commencing a main randomised controlled trial (RCT) [17,18,19]. Feasibility studies are designed to answer whether the study protocol can work and allows for modification of the protocol before commencement of the main trial [19, 21,22,23]. Hence, the aims of this study were to assess the feasibility of the data collection procedures and the acceptability of the Ad-Shoulder intervention.

Methods

Trial design

The feasibility study had a single-group pre-post intervention design. The trial is reported according to the Consolidated Standards of Reporting Trials (CONSORT) 2010 statement: extension for pilot/feasibility studies (Additional file 1) [24]. The trial protocol was registered with the ClinicalTrials.gov registry (NCT04190836, December 9, 2019) and approved by the Norwegian Regional Ethical Review Board (ref. no. 2017/355, April 21, 2017).

Participants

The inclusion criteria were as follows:

  • Adults (aged > 18 years) with shoulder pain located in the upper arm

  • Previously received conservative care due to subacromial pain but still seeking primary or secondary care during the past 6 months

The exclusion criteria were as follows:

  • Bilateral shoulder pain

  • Clinical presentation consistent with a frozen shoulder diagnosis (< 50% external rotation compared to contralateral side) [8].

  • Patients who have received surgical treatment due to shoulder problems

  • Pregnancy

  • Patients with insufficient Norwegian language skills

  • Serious psychiatric disorder

Recruitment and enrolment

Recruitment for this feasibility study was conducted by two practitioners (1 general practitioner and one physiotherapist) across two sites from November 2017 to January 2018 and May 2018. The two outpatient clinics were located in Oslo, Norway, of which one in primary care (Oslo Metropolitan University) and the other in secondary care (Diakonhjemmet Hospital). Potential participants were identified when seeking care for shoulder pain. The potential participants were pre-screened by a physiotherapist or a medical doctor at one of the recruitment sites. The final enrolment was conducted by two researchers (DHM and HVG) that would be providing the intervention, and eligibility for the study was re-checked. Informed consent was provided by all participants at inclusion after being provided oral and written information. During the recruitment process, it was deemed necessary to include additional eligibility criteria because the physiotherapists conducting the assessment at the second screening diagnosed patients with other diagnosis than subacromial pain. Therefore, clinical signs of a total rotator cuff tear, clinical signs of instability, clinical signs of a cervical syndrome, clinical signs of AC joint arthritis and reasons to suspect systemic pathology including inflammatory disorders were added to the exclusion criteria [8]. Clinical signs of subacromial pain was added to the inclusion criteria to confirm the diagnosis [8].These criteria are consistent with the British Elbow and Shoulder Society (BESS) guidelines [8].

The Ad-Shoulder intervention

The Ad-Shoulder intervention was developed by DHM and YR on the basis of the self-managed single exercise programme by Littlewood et al. [25,26,27,28,29] and informed by recent research on subacromial shoulder pain [10, 12] and adherence to exercises for patients with persistent musculoskeletal pain [30]. The behavioural component of the Ad-Shoulder intervention was based on the self-management framework, provided by Lorig and Holman [31], A key component in this framework is to target patients’ self-efficacy, defined as the confidence to perform a specific task or behaviour [32]. In Bandura’s Social Cognitive Theory, the person’s perceived self-efficacy is thought to mediate behaviour change [32], which in this self-management intervention is closely linked to adherence to home exercises and physical activity. In the self-management framework suggested by Lorig and Holman [31], the ability to self-manage is achieved based on learning five core self-management skills. These are problem-solving, decision-making, resource utilisation, the forming of a patient/health care provider partnership and taking action. These core self-management skills are elaborated according to the intervention in Appendix 1. The intervention consisted of 1–5 individual sessions over 3 months, where the first session had a duration of 1 h and the following sessions about 45 min. The self-management strategy emphasises dynamic, progressively loaded exercises for the shoulder (Table 4). To enhance exercise adherence, we used behaviour change techniques such as social support, goal setting, demonstration of behaviour, graded tasks and self-monitoring of behaviour (Appendix 1). For specific content reporting of the home based exercises, we have followed the Certificate on Exercise Reporting Template (Additional file 2) [33, 34]. The participants had the option to contact the physiotherapist by phone, text message or e-mail for advice for up to 12 weeks, the duration of the intervention. Patients were allowed to continue with their usual medication, but were asked not to receive other treatment. The intervention was delivered by a PhD student (DHM) and a master student (HVG), both qualified physiotherapists’, with a special interest in shoulder pain and at least 4 years’ experience with assessment and management of musculoskeletal pain conditions.

Measurements

The participants filled in a self-reported questionnaire at baseline (right before the first consultation) and at week 12. In addition, three self-reported measures (Pain Self-Efficacy Questionnaire 2-item, Numeric Pain Rating Scale and Self-Efficacy) were collected repeatedly, and data on objective physical activity was collected with accelerometers one week before baseline, at week 6 and at week 12.

The questionnaire package consisted of sociodemographic variables (age, sex, duration of shoulder pain, education level, work status, relationship status, smoking status, height and weight) and patient-reported outcome measures. An overview of the self-reported measures and time points is provided in Table 1 and a description of these are provided in Appendix 3.

Table 1 Overview of self-reported measures and the time points for the measurements
Full size table

The participants also responded to SMS text messages containing three measurements (Numeric Pain Rating Scale, Self-efficacy and Pain Self-Efficacy Questionnaire 2-item) before, during and after the intervention period to explore the process of change on an individual level. During the first week pre-treatment phase (A), 3 text messages including the three measurements were collected. During the 12-week treatment phase (B) the text messages were collected twice every week (total of 24), and during the one week post-treatment phase (A) the text messages were collected 3 times.

Physical activity was objectively measured, using four accelerometers (AX3, 3-axis Logging Accelerometer, Axivity, UK) attached to the chest, upper arm, hip and wrist for 3 consecutive days 1 week before baseline, at week 6 and at week 12. This accelerometer provides information about movement and enabled us to objectively measure the amount of general physical activity (minutes of ≥ moderate activity, defined as metabolic equivalent (MET) value ≥ 3).

Feasibility outcomes and progression criteria

  1. 1.

    The recruitment rate—measured by how many people that were eligible and how many people that were recruited per week

Progression criteria: Ten participants would be enrolled in the study within a 12 week period.

  1. 2.

    Follow-up rate—measured by the percentage of participants who were followed up successfully until the three months follow-up

Progression criteria: Follow-up rates ≥ 80% for the questionnaire at 12 weeks and the repeated measures.

  1. 3.

    Feasibility of actigraph assessment of physical activity—measured by percentage of participants with valid data at each time point (baseline, 6 weeks and 12 months). Validity of data was defined as successful measurement of a participant’s physical activity for at least 20 h day for three consecutive days.

Progression criteria: At least 80% of participants would contribute valid physical activity (actigraph) data at each time point.

  1. 4.

    Adherence with the self-managed exercises—measured by percentage of patients maintaining at least 80% adherence to the self-managed exercises measured by self-reported exercise logbook.

Progression criteria: At least 80% of the participants would adhere to at least 80% of the assigned home exercise programme (self-reported).

  1. 5.

    Fidelity of the delivery of the intervention—measured by whether the physiotherapist delivered the components of the intervention or added other components to the intervention using a physiotherapist completed logbook at 3 months.

Progression criteria: The therapists would deliver the intervention according to the protocol (100%).

  1. 6.

    Number and nature of adverse events—measured by self-report questionnaire.

Progression criteria: No more than 30% would report adverse events (including mild), such as increased short term pain with home exercises.

  1. 7.

    Data collection procedure—assessed by exploring data from the self-report outcomes with respect to missing data and scoring pattern at baseline (floor/ceiling effect, median, variation).

In terms of decision-making against the progression criteria, should we have fallen below/above any of these rates in our feasibility study we would consider whether protocol modification or close monitoring during a main RCT would address any failure to meet these criteria, or decide that the main RCT would not be feasible [19].

Sample size

As this is a feasibility study and inferential statistics were not calculated there was no need to achieve a desired power to detect an effect. The research team decided that 10 participants would be sufficient to give a preliminary understanding of the feasibility of data collection procedures and acceptability of the Ad-Shoulder intervention.

Blinding

Due to practical reasons, the patients and clinicians were not blinded to the treatment and the researcher conducting the analyses (DHM) was one of the therapists in the intervention group.

Statistical methods

Descriptive statistics were used to assess the feasibility objectives and the patient-reported outcomes using SPSS (version 25, IBM, Armonk, NY, USA) and Microsoft Excel (2016). Due to the low number of participants, the continuous variables were presented as median and interquartile range or min–max values.

Results

Sample characteristics

The included participants had a median age of 48 years (IQR 15), seven of the 11 participants were female, and the median duration of shoulder pain was 18 (IQR 15) months. The baseline characteristics of participants are provided in Table 2.

Table 2 Patient demographics and patient characteristics (n = 11)
Full size table

Recruitment capability

Flow of participants through the study and reason for exclusion is presented in Fig. 1. The 11 patients were included during a time-period of 16 weeks (November 2017–January 2018 and May 2018). We had to use 4 more weeks than what was anticipated before commencement of the study.

Fig. 1
figure 1

CONSORT 2010 flow diagram. Design and flow of participants through study

Full size image

Follow-up

Ten of the eleven participants completed the questionnaire package at baseline and week 12. For the repeated measures, 313 of 330 (95%) were collected. There were 17 occasions of a reminder being sent; of these, all except two were successful. Ten of the eleven included patients delivered complete exercise logbooks at week 6 and week 12.

Objective assessment of physical activity

Concerning the actigraph measurements, objective physical activity was successfully obtained for 100% (11/11) of the participants at baseline, 64% (7/11) participants at week 6 and 82% (9/11) participants at week 12. The reasons for the missing data were time constraints (one participants at week 6), being abroad on holiday (one participant at week 6), skin eczema (one participant at week 6 and 12), skin irritation (one participant at week 6) and one patient dropped out of the study (one participant at week 12). The calculated time-consumption for administration of the four actigraph sensors was a total of 1 h per patient at each time point, excluding data analysis. In the analysis of the data, we were able to obtain a good impression of the general physical activity level, using only one accelerometer (wristband).

Adherence

The included patients reported a median of 3 sessions (range 1–5) during the 12-week intervention period. With respect to treatment adherence, all patients met to the scheduled appointments.

Based on information obtained from the exercise log, median adherence was 86.5% (range 47–97). Fifty-five percent (6/11) of the participants satisfactorily completed at least 80% of their home exercise programme. The four patients that had self-reported adherence to exercises < 80% reported the main barriers to adherence to be time constrains (n = 2), forgetfulness (n = 1) and one patient was not able to do the exercises due to periods of sickness (n = 1).

Fidelity

According to the logbook completed by the physiotherapists that delivered the Ad-Shoulder intervention, all sessions were delivered as planned. One of the therapists (HVG) contacted one of the developers of the intervention (DHM) twice via telephone to get advice and to make sure she adhered to the treatment protocol.

Adverse events and other treatment

None of the patients reported any adverse events at 12 weeks follow-up. One patient (case 4 see Fig. 2) reported that she had received other treatments during the intervention period (manual therapy and one cortisone injection).

Fig. 2
figure 2

Repeated measures at 30 occasions for the 11 patients converted to a value of 0–100, where 100 indicates a better score. Measurement number 1–3 were 1 week before the intervention started, 4–27 were done twice weekly during the intervention period (12 weeks), and 28–30 were 1 week after the intervention. NPRS, Numeric pain rating scale; PSEQ2, Pain Self-Efficacy Questionnaire 2 item; Self-efficacy, “How confident have you felt about managing your shoulder pain by yourself?”

Full size image

Patient-reported measurements

The results of the patient-reported measures are displayed in Table 3. The median change from baseline to week 12 in the SPADI was 18 (range − 0.8 to 53). The data from the repeated measures (NPRS, PSEQ-2 and general self-efficacy) are presented for every individual and the group in Fig. 2. At baseline the median PSEQ-2 score was 9 (range 7–12), where three of the 11 patients scored the maximum score. Five of the included patients felt very confident in being able to manage their shoulder pain by themselves at baseline (general self-efficacy). The median change from baseline to week 12 in PSEQ-2 and general self-efficacy was 0 (range − 1 to 3) and 1 (range 0–3), respectively.

Table 3 Measurements at baseline and 12 weeks follow-up (n = 11)
Full size table

Discussion

The present feasibility study demonstrated that the Ad-Shoulder protocol was feasible with respect to follow-up rate and objective physical activity measurements at baseline and week 12. Furthermore, the Ad-Shoulder intervention was acceptable in terms of intervention fidelity and no adverse events were reported. In contrast, the recruitment rate and the patients’ adherence to the self-managed exercise programme were below our predefined progression criteria.

The Ad-Shoulder intervention was designed to enhance adherence to the self-managed exercises through a behavioural framework. One of the main challenges in light of progressing to a main RCT was that only six of 11 participants had an acceptable adherence rate to self-managed home exercises, despite that they had excellent attendance to physiotherapy sessions and that the physiotherapists had good fidelity to the Ad-Shoulder protocol. Our finding suggests that additional initiatives need to be taken in order to enhance adherence to the self-managed exercise programme between the consultations with the physiotherapist(s). Qualitative research to explore barriers and facilitators to adherence to the self-managed exercises might help the research team to develop strategies to enhance the adherence rate before commencing on future clinical trials [49, 50].

The lack of recruitment capability in the present study provides another challenge in the perspective of conducting the main RCT. The reason for the low recruitment rate was due to a low number of eligible participants. However, all eligible participants consented to participate in our study. This is similar to a Norwegian RCT among patients with subacromial pain, where only two out of 141 (1.5%) declined to participate [51]. In a RCT with a calculated sample of approximately 150 patients, the present results indicate a recruitment period of 218 weeks (over 4 years) and screening of approximately 300 potentially eligible patients. To be able to conduct a large multicentre trial with several involved people, we will require additional funding for necessary staff involved in the different phases of the trial (recruitment, randomisation, delivery of the treatment, follow-up procedures and analyses).

The change in the clinical outcomes during the 12 weeks of follow-up demonstrated that most participants improved. The median change on the SPADI of 17 points is slightly above the estimate for a minimal important change, which has been reported to be 8–13.2 [52]. However, it is important to interpret these findings carefully, as we cannot distinguish between effects that might have occurred due to the natural course of the condition, regression to the mean, placebo or the Ad-Shoulder intervention. Similar improvements were seen in pain intensity during the 12 weeks of follow-up, whereas in the two potential mediator measurements (self-efficacy and PSEQ-2) for the future RCT, little change occurred. The lack of change in pain self-efficacy and general self-efficacy might be due to the particular measurements we used, due to the high scores among many of the included patients at baseline or that the intervention is not performing as we assumed that it would. This finding warrants further exploration of whether pain self-efficacy and general self-efficacy are the most relevant mediators to explore in a future main RCT.

Strengths and limitations

A strength of this study is that we used important benchmarks that have previously been recommended to assess the feasibility and acceptability of the trial [19, 21,22,23]. Furthermore, we used repeated measures of core outcome constructs in order to explore whether we can expect clinically relevant improvements in these measurements and included patient-reported outcome measures at baseline and week 12 in line with the core domain set for clinical trials of shoulder disorders recommended by the OMERACT group [53].

The study has limitations. First, because of the limited number of participants enrolled in this study it is difficult to determine whether the estimated rates addressing the feasibility outcomes (e.g. recruitment rate) are representative for the targeted population. Secondly, adherence to the self-managed exercises was lower than anticipated and because we did not include qualitative interviews, methods were not in place to explore the reasons in depth. A third limitation is that fidelity was measured in a limited way using only the physiotherapists’ journal notes. The intervention was delivered by experienced physiotherapists, which might have ensured high fidelity to the Ad-Shoulder protocol. Whether the fidelity will be equally high when implemented to other settings, for example in a multicentre trial, cannot be predicted from the present results. A fourth limitation is that we did not assess follow-up in a 12-month perspective as proposed for the main RCT. Therefore, the 3 months follow-up rate might be a bit optimistic of what we can expect at 12 months follow-up. Finally, we were not able to measure home exercise adherence objectively during the 12 weeks intervention period and the self-reported adherence might therefore be overestimated.

Conclusions

The feasibility of recruitment and adherence to the self-managed exercise programme were both below the anticipated level. Valid objective physical activity data measures at baseline and week 12, follow-up methods, the physiotherapists’ fidelity to the intervention and patients reporting adverse events were all acceptable according to our progression criteria. Additional recruitment sites will be added to optimise recruitment and further intervention development including qualitative research is necessary to understand whether adherence to the self-managed exercise could be enhanced.