Anti-TNF (adalimumab) injection for the treatment of pain-predominant early-stage frozen shoulder: the Anti-Freaze-Feasibility randomised controlled trial

Objective The Anti-Freaze-F (AFF) trial assessed the feasibility of conducting a definitive trial to determine whether intra-articular injection of adalimumab can reduce pain and improve function in people with pain-predominant early-stage frozen shoulder. Design Multicentre, randomised feasibility trial, with embedded qualitative study. Setting Four UK National Health Service (NHS) musculoskeletal and related physiotherapy services. Participants Adults ≥18 years with new episode of shoulder pain attributable to early-stage frozen shoulder. Interventions Participants were randomised (centralised computer generated 1:1 allocation) to either ultrasound-guided intra-articular injection of: (1) adalimumab (160 mg) or (2) placebo (saline (0.9% sodium chloride)). Participants and outcome assessors were blinded to treatment allocation. Second injection of allocated treatment (adalimumab 80 mg) or equivalent placebo was administered 2–3 weeks later. Primary feasibility objectives (1) Ability to screen and identify participants; (2) willingness of eligible participants to consent and be randomised; (3) practicalities of delivering the intervention; (4) SD of the Shoulder Pain and Disability Index (SPADI) score and attrition rate at 3 months. Results Between 31 May 2022 and 7 February 2023, 156 patients were screened of whom 39 (25%) were eligible. The main reasons for ineligibility were other shoulder disorder (38.5%; n=45/117) or no longer in pain-predominant frozen shoulder (33.3%; n=39/117). Of the 39 eligible patients, nine (23.1%) consented to be randomised (adalimumab n=4; placebo n=5). The main reason patients declined was because they preferred receiving steroid injection (n=13). All participants received treatment as allocated. The mean time from randomisation to first injection was 12.3 (adalimumab) and 7.2 days (placebo). Completion rates for patient-reported and clinician-assessed outcomes were 100%. Conclusion This study demonstrated that current NHS musculoskeletal physiotherapy settings yielded only small numbers of participants, too few to make a trial viable. This was because many patients had passed the early stage of frozen shoulder or had already formulated a preference for treatment. Trial registration number ISRCTN 27075727, EudraCT 2021-03509-23, ClinicalTrials.gov NCT05299242 (REC 21/NE/0214).


INTRODUCTION
Frozen shoulder (adhesive capsulitis) is a common and extremely painful and debilitating condition.Affected individuals struggle with activities of daily living and significant sleep disturbances as a result of severe pain. 1

STRENGTHS AND LIMITATIONS OF THIS STUDY
⇒ The Anti-Freaze-F study was a pilot randomised controlled trial with an embedded qualitative study to test the effectiveness of a new treatment for people with early stage, pain-predominant frozen shoulder.⇒ Participants were randomised (centralised computer generated 1:1 allocation) to either ultrasoundguided intra-articular injection of adalimumab or placebo injection.Participants and outcome assessors were blinded to treatment allocation.⇒ We assessed the feasibility of conducting a large multicentre randomised trial in terms of the ability to screen and identify participants, their willingness to consent and be randomised, the practicalities of delivering the intervention and attrition rates at 3 months.⇒ While all nine participants recruited into the trial received both their first and second injections within the specific timeframe, it is unclear whether this would be same for larger definitive trial.⇒ Similarly, while all participants completed both patient-reported and clinician-assessed outcomes at 3 months, it is unlikely this would be same for larger trial with longer follow-up.

Open access
It affects about 9% of people in the United Kingdom (UK) aged 25-64 years, 2 and 20% develop the same problem in the other shoulder. 3Frozen shoulder may develop as a primary condition or secondarily following surgery or trauma.
The classic description of the development of frozen shoulder is of three overlapping phases. 4 5The initial pain-predominant inflammatory phase is characterised by constant pain and difficulty sleeping and lasts between 3 and 9 months.This progresses to a stiffnesspredominant fibrotic phase, with progressive restriction of motion, particularly external rotation and elevation of the shoulder, and impairment of function and lasts between 4 and 12 months.The pain changes from being constant to being manifest at the end of range of motion and of reduced intensity.Over the final phase, there is a gradual improvement in range of motion and stiffness over a 12 to 48-month period, although pain may persist with end-range movements.The average duration of the condition is 30 months (range 1-3.5 years) although full resolution of symptoms does not always occur. 6he aetiology of frozen shoulder is poorly understood and consequently there is no consensus on the optimal treatment.The majority of patients with early-stage painpredominant frozen shoulder are managed in primary care or at primary care interface musculoskeletal services by physiotherapists and general practitioners.During this stage, standard treatment consists of rest, advice, analgesics, physiotherapy and corticosteroid injections to address the symptoms.Current treatments available to patients with frozen shoulder are of limited efficacy.Two Cochrane reviews concluded that while oral steroid or local steroid injections lead to short-term benefit in pain and range of motion, the effects are not maintained beyond 6 weeks. 7 8A more recent systematic review (based on data from five randomised trials) also found some benefit of corticosteroid injection compared with placebo, but this pain relief was not sustained in the long term. 9Other Cochrane reviews concluded that there is no evidence that physiotherapy or ultrasound therapy are beneficial, 10 and that manual therapy with exercise is less effective than corticosteroid injection in the short term. 11he UK FROST compared the effects of physiotherapy plus corticosteroid injection, manipulation under anaesthesia with a steroid injection, and arthroscopic capsular release supplemented with manipulation. 12None of the treatments was found to be clinically superior.
Our study (Anti-Freaze-F) was designed to specifically target people with early-stage pain-predominant frozen shoulder.We assessed the feasibility of conducting a large multicentre randomised trial to test whether giving an intra-articular injection of adalimumab (a drug targeting the inflammatory mediator tumour necrosis factor (TNF)) can reduce pain and prevent the condition from getting worse, if given during the early pain-predominant stage.
Approximately 50% of people with frozen shoulder also develop Dupuytren's disease. 13Dupuytren's disease has a prominent genetic component, and in a study of twins, the heritability was estimated at 80%. 14 We have recently confirmed that there is a significant genetic correlation between Dupuytren's disease and frozen shoulder. 15he pathology of the two conditions has several similarities. 16 17A systematic review of the pathophysiology of frozen shoulder identified the presence of fibrosis and the role of inflammation. 6The affected tissues are infiltrated by immune cells and there are elevated levels of proinflammatory cytokines, including TNF, 18 19 with myofibroblasts driving the development of fibrosis. 16In a dose-ranging proof of concept phase 2a followed by a 2b clinical trial (RIDD trial), we found that injection of anti-TNF (adalimumab) directly into Dupuytren's nodules led to significant downregulation of the myofibroblast phenotype and reduction in nodule hardness (primary endpoint) and nodule size on ultrasound scan (secondary endpoint). 20 21

Objectives
The aim of the Anti-Freaze-F (AFF) trial was to assess the feasibility of conducting a large randomised controlled trial to assess whether an intra-articular injection of adalimumab (anti-TNF) can reduce pain and improve function in people with early-stage frozen shoulder.The primary objectives were to assess the: ► Ability to screen and identify potential participants with pain-predominant early stage frozen shoulder.► Willingness of eligible participants to consent and be randomised to intervention.► Practicalities of delivering the intervention, including time to first injection and number of participants receiving second injection.► SD of the Shoulder Pain and Disability Index (SPADI) score and attrition rate at 3 months (ie, 12 weeks) postrandomisation in order to estimate the sample size for a definitive trial.Secondary objectives were to assess the follow-up rates and viability of patient-reported outcome measures and clinician-assessed shoulder range of motion at 3 months (ie, 12 weeks).
The objectives of the embedded qualitative study were to explore the participant experience of being recruited to the AFF trial, the treatment received and follow-up schedule, and to understand what helps participant recruitment to the trial intervention.

Study design
A multi-centre, randomised, double-blind, parallel group, feasibility trial with an embedded qualitative study.

Setting
Participants were recruited from four National Health Service (NHS) musculoskeletal services and their related physiotherapy services, with treatment delivered within Open access these services or local secondary care dependant on the local service provision.

Study participants
Participants were eligible if they were aged 18 years and above with a new episode of shoulder pain attributable to pain-predominant stage of frozen shoulder (ie, within approximately 3 to 9 months of onset of symptoms), which was diagnosed clinically using the criteria set out in the British Elbow and Shoulder Society (BESS) guidelines. 22In line with the trial protocol, imaging, including plain radiographs, may be used to help confirm a diagnosis of frozen shoulder by excluding other pathology such as glenohumeral arthritis as per standard NHS care.However, is not always necessary and hence routine radiology was not as part of the trial procedures. 23Patients were excluded if they had frozen shoulder secondary to significant shoulder trauma; other shoulder disorders (eg, inflammatory arthritis, rotator cuff disorders, glenohumeral joint instability) or with red flags consistent with the criteria set out in the BESS guidelines 22 ; bilateral early stage frozen shoulder; had received corticosteroid injection for shoulder pain in the last 12 weeks; were currently taking any anti-TNF drug, or being treated with coumarin anticoagulants, such as warfarin; had significant renal or hepatic impairment, or had contraindications to anti-TNF injection.Detailed inclusion and exclusion criteria are described in the trial protocol. 23atients who met the eligibility criteria and wanted to participate were approached for written informed consent by a research facilitator trained in Good Clinical Practice at each participating site.All participants underwent serological testing to check for latent tuberculosis (TB) and hepatitis B surface antigen.The blood tests were performed during the baseline assessment or at the time the participant attended for their first injection appointment (depending on the local site provision).The risks of reactivation following a single injection are low.If any participants had shown a positive serology test result, they would have been referred to their local infectious disease service and would not have received the second injection.

Randomisation
Consented participants were randomly assigned to receive either: (1) intra-articular injection of adalimumab or (2) placebo injection (saline (0.9% sodium chloride), both under ultrasound guidance.Randomisation (1:1) was done by the research facilitator at each site using the centralised randomisation service provided by the Oxford Clinical Trials Research Unit once the patient was enrolled and baseline assessment and questionnaire were completed.Randomisation was computer generated and stratified by study site using a variable block size (variable block sizes of 2, 4 and 6 in a ratio of 1:2:1) to ensure the participants from each study site had equal chance of receiving either intervention.

Blinding
Study participants and site staff, except pharmacy staff, were blinded to treatment allocation.The clinician delivering the treatment injection was not blinded but was not involved in further trial-specific assessment of the participant.The trial statistician and data entry personnel were not blinded to the treatment allocation.The remaining members of the trial management team, including the staff conducting the qualitative interviews, were blinded to treatment allocation until after data analysis was complete.

Adalimumab/placebo injection
As per their treatment allocation, participants received either intra-articular injection of adalimumab (160 mg in 3.2 mL for the first injection, 80 mg in 1.6 mL for second injection) or placebo (normal saline (0.9% sodium chloride) 3.2 mL for the first injection and 1.6 mL for the second injection).Full details of the injection delivery have been reported previously 23 and are described here in brief.
Appointments were coordinated, so that participants received their first injection within approximately 2 weeks of randomisation, and second injection 2-3 weeks after the first (unless the participant declined, tested positive for TB or hepatitis B surface antigen or had a related grade 3 or above adverse event).The injection was given into the anterior shoulder joint space in the rotator cuff interval where there is maximal inflammation of the capsule and synovium, 24 under guided ultrasound by an appropriately qualified practitioner.
The adalimumab/placebo injection was dispensed by the local site pharmacy and sealed in identical sized and sealed opaque packaging.Preparation of the adalimumab/placebo injection took place in a clinic room/ area separate from the participant to ensure the participant remained blinded to treatment allocation.Both adalimumab and placebo have a similar appearance; therefore, the two treatments are indistinguishable.The skin at the site of injection of adalimumab/placebo was infiltrated with local anaesthetic to reduce the pain of the injection as per local practice.Injection details, including success of the intra-articular injection (ie, fully or partial), were recorded on a trial-specific injection treatment log.
All participants received a written physiotherapy advice leaflet providing education and advice about frozen shoulder and pain management. 4The advice leaflet also included simple self-guided exercises, which participants could use to increase shoulder joint movements, once the early pain-predominant stage reduced.Participants were advised they may seek other forms of treatment during the follow-up period of the trial but were informed that they should use usual routes (eg, through GP referral) to do so.Participants were also advised that they could seek corticosteroid injection but were asked to wait until after the 3-month follow-up period.

Outcome measures
Feasibility success criteria To determine the feasibility of a definitive randomised controlled trial, the prespecified success criteria were: ► Feasibility to recruit: ≥33% of potentially eligible patients with frozen shoulder screened eligible for recruitment.► Success of consent process: ≥33% of eligible participants consented.► Intervention delivery: ≥75% of participants receive first injection as randomised within specified timeframe.In addition, we collected outcomes at baseline and at 3 months to assess the feasibility of collecting these in a future definitive trial and to obtain the variability estimates required for estimation of the sample size of a definitive trial.Patient-reported outcomes, collected via online/paper questionnaire, included shoulder pain and function measured using the SPADI scale (primary outcome for definitive trial); 25 26 subdomains of pain (SPADI 5-item pain subscale), function (SPADI 8-item disability subscale) 25 26 ; shoulder range of motion (Participant Shoulder Movement Questionnaire); psychological factors (Fear-Avoidance Belief Questionnaire) 27 ; pain self-efficacy questionnaire 28 ; sleep disturbance (Insomnia Severity Index) 29 ; patient global impression of change 30 ; return to desired activities; and additional health resource use.At baseline and the 3-month face-toface follow-up appointment, a blinded assessor measured the shoulder movements including active flexion, extension, abduction internal and external rotations, using a manual goniometer.

Adverse events
The safety profile of adalimumab is well known, with the most common adverse reactions being mild injection site reactions.The Common Terminology Criteria for Adverse Events (CTCAE) V.5.0 was used to guide recording of any adverse events including grading of the event.Only clinician-assessed adverse events, graded 3 and above, occurring during the trial for each participant, from their consent until the 3-month follow-up, related to the trial medication (adalimumab/placebo) were recorded.

Sample size
The main feasibility objective and, therefore, the basis of the sample size estimate were participant recruitment per centre.The target sample size was 84 participants, equivalent of 1 to 2 participants per month per site over 12 months.Seventy is the recommended minimum target sample size when including an estimate of the SD in an external pilot trial. 31The sample size was increased to 84 to increase precision of the estimate of the SD of SPADI at 3 months, the proposed primary outcome for the definitive trial and to consider possible attrition based on an attrition rate of 15%.

Statistical analysis
Feasibility outcomes were reported as numbers and percentages.Baseline characteristics and outcome measures were reported using descriptive statistics, separately per group and overall, using either the mean and SD for continuous variables and number and percentage of participants in each group for categorical variables.

Embedded qualitative study
We planned to interview a purposive sample of up to 15 participants (or until we reach data saturation) to provide variability for age, gender, ethnicity and geographical representation.A qualitative researcher (CS), blinded to treatment allocation, conducted the telephone interviews using a semistructured interview guide with open-ended questions.Interviews were audio-taped and transcribed verbatim.Data were analysed following Braun and Clarke's thematic analysis method. 32Interviews were coded using NVivo qualitative software V.12 and categorised into themes.Participant narrative quotes were used to illustrate the themes.

Patient and public involvement
Patient representatives were involved in the design, conduct and reporting of study, including reviewing of patient-facing study materials (eg, patient information sheet, physiotherapy advice booklet and patient follow-up questionnaires) and through involvement of the trial management and trial oversight committees.

RESULTS
Regulatory approvals were obtained on 22 December 2021.Patient screening and enrolment began 31 May 2022 and ended 7 February 2023; completion of the 3-month outcome assessment and follow-up ended 9 May 2023.Delays in obtaining site approvals due to the continuing impact of COVID-19 on staffing levels meant that site opening and thus patient screening was delayed, resulting in the recruitment period being just over 8 months as opposed to the planned 12 months.Delays in local approvals at sites also meant we were only able to open four of the planned five sites as stipulated in the trial protocol.

Feasibility objectives Feasibility to recruit
Of the 156 patients screened, 39 (25%) were eligible for the trial.The main reasons screened patients were not eligible were because they had other shoulder disorders (eg, inflammatory arthritis, rotator cuff disorders, glenohumeral joint instability) (n=45/117; 38.5%); were no longer in the pain-predominant stage of frozen shoulder (n=39/117; 33.3%); had frozen shoulder due to significant shoulder trauma or other causes (eg, recent breast cancer surgery) (n=11/117; 9.4%); had bilateral early stage frozen shoulder (n=7/117; 6.0%) or had received Open access corticosteroid injection for shoulder pain within the last 12 weeks to either shoulder (n=5/117; 4.3%) (figure 1).

Success of consent process
Of the 39 patients eligible for the trial, 29 (74.4%) declined to participate, the main reasons being they already had a treatment preference for receiving steroid injection (n=13/29; 44.8%), did not wish to be randomised (n=7/29; 24.1%) or already had a treatment preference for not receiving an injection (n=4/29; 13.8%).In addition, one patient who was eligible was not consented due to close of trial recruitment.This resulted in nine (n=9/39; 23.1%) eligible patients being consented into the trial (figure 1); four were randomised to receive adalimumab injection and five to placebo injection.Baseline characteristics are summarised in table 1. Participants had a mean age of 54.9 (SD 5.6) years and three (33.3%) of the nine participants were women; all were white British.The average (mean) duration of symptoms was 15.4 weeks (SD 9.0).The mean SPADI score at baseline (primary outcome for definitive trial) was 70.8 (SD 16.6) for the adalimumab group and 55.4 (SD 21.5) for the placebo group.

Intervention delivery
All participants received their treatment as allocated and in line with the trial protocol.The mean time from randomisation to first injection was 12.3 days for the adalimumab group and 7.2 days for the placebo group.One participant in the adalimumab group received their injection 18 days post-randomisation as they were not available until then.The mean time from receiving first to second injection was similar for adalimumab (18.5 days) and placebo (20.8

days) (table 2).
Patient retention and outcome measure data collection Data for patient-reported outcomes and clinicianassessed range of shoulder motions are shown in online supplemental tables S1-S3.The completion rates at baseline and 3 months were 100% (n=9/9) for both patient-reported outcomes, completed by questionnaire, and clinician-assessed shoulder motion.Between-group difference were not calculated as only nine participants were recruited into the trial.Online supplemental figure S1 shows the change from baseline and at 3 months for shoulder pain and function (measured using the SPADI scale) per participant for adalimumab and placebo injection.

Adverse events
No graded 3 or above, adverse events occurred during the trial related to the trial medication (adalimumab/ placebo).

Embedded qualitative study
Eight participants took part in the telephone interviews (four from each injection group), which included six men and two women with an average age of 55.5 years (range 46 to 62 years).The average duration of interviews was 24 min (range 14 min-31 min).Five participants had received both injections at the time of interviews.Four key themes emerged: (1) experiences with pain-predominant stage of frozen shoulder; (2) perceptions about participating in the AFF trial; (3) perceptions about the trial processes and (4) perceptions about the effects of injections.Themes and illustrative quotes are presented in table 3.
Experiences with pain-predominant phase of frozen shoulder This theme describes the impact of the pain-predominant phase of the frozen shoulder.Most participants reported severe levels of pain, particularly with certain movements or during sleep.The shoulder pain was described as 'very acute', 'annoying', 'disruptive', 'excruciating' or 'completely unbearable'.Participants also discussed how their everyday function was limited (eg, difficulties with getting dressed or driving).Sleep was substantially affected by pain, with participants reporting shorter sleep times and poorer sleep quality.They frequently woke up in the night due to severe pain while changing position.

Perceptions about participating in the trial
This theme describes the participants' views on participating in the AFF trial.Overall, participants were keen to find a solution to their shoulder problem.Severe pain, reduced function and sleep and previous treatments such as steroid injections or physiotherapy that were not helpful were other reasons that influenced their decision to take part.Participants also reported altruistic reasons to benefit others with similar problems.Their expectations on taking part were to get adalimumab injection, pain relief and improved function.Participants also expressed a clear understanding and acceptance of the randomisation process, mainly attributing to the trial-related information provided to them.Having a 50% chance of getting the placebo injection was a concern.However, participants discussed various reasons, including long waiting time for steroid injections, positive expectations about adalimumab and flexibility to access standard care (if AFF trial injections did not work) that helped them take a 'balanced view' in participating.

Perceptions about trial processes
This theme describes participant views about trial processes, including provision of information, delivery of injections, outcome assessments and their overall experience.Participants described that all trial-related information such as randomisation and delivery of injections was communicated effectively, with opportunities to clarify their queries and make a decision.Injection appointments were scheduled quickly, and participants were well-informed and supported during the sessions.Many reported discomfort at the time of injections.Participants perceived the outcome measurements (clinician-assessed measurements and self-reports) as relevant to their condition and had no issues completing them.They also highlighted their positive experiences with the trial personnel Open access at study sites and reported an overall positive experience of participating in the AFF trial.

Perceived effects of adalimumab and placebo injections
Two participants who received adalimumab injections reported improvements in pain and/or shoulder movements.Another participant reported some pain relief after the first injection.Pain remained unchanged and movement decreased after the second injection.One participant who had received first injection only reported improvement in movement.Two participants who received placebo injections reported increased pain after their first injection that persisted after the second injection.Of the two participants who had only their first injection, one reported noticeable improvement in pain and movement; the other reported increased pain.Two participants (one from each group) were concerned about the increased pain after the injections.Other concerns were expectations about the long-term effects of adalimumab and continued access to physiotherapy within the trial.

DISCUSSION
Our findings show that a definitive trial to assess whether an intra-articular injection of adalimumab (anti-TNF) can reduce pain and improve function in people with earlystage frozen shoulder was not feasible within the NHS musculoskeletal and related physiotherapy services that participated in the AFF trial.Only a quarter of the people screened were eligible to take part in this feasibility study; of those eligible, only nine consented to be randomised.Key themes from the semistructured interviews provided a useful insight into the trial and focused on the participant experiences during the pain-dominant stage of frozen shoulder and their perceptions about trial participation, trial processes and outcomes following injection.
One of the main reasons people were ineligible to take part in the AFF trial was because they were no longer in the pain-predominant phase of frozen shoulder by the time they were seen by the physiotherapist.People no longer being in the pain-predominant phase of frozen shoulder is indicative of long waiting lists and time taken to be seen across NHS musculoskeletal and related physiotherapy services and was a major barrier in terms of trial feasibility.Long waiting times is a problem across the NHS, particularly during and after the COVID-19 pandemic; 33 the period in which the AFF trial was recruiting and not specific to people with frozen shoulder.The four NHS sites that participated in the AFF trial represented several different musculoskeletal and physiotherapy service delivery approaches, with approaches also changing during and post the COVID-19 pandemic.Several of these sites also included a small number of First Contact Practitioners (FCPs) embedded with primary care.Open access shoulder community and created a short trial promotional video for the Chartered Society of Physiotherapists.We also promoted the trial on social media with input from our patient representative, and through patient organisations such as the Dupuytren's Society and Diabetes UK to target groups in whom the incidence of frozen shoulder is higher. 1 13 Another main barrier to feasibility of the trial was because some people, who were eligible and approached to take part in the AFF trial, already had a treatment preference for receiving a steroid injection and declined to participate.We did consider the use of corticosteroid injection as opposed to placebo when designing the trial; however, given the available evidence showing that steroid injection only leads to short-term benefit, 8 9 we decided to use a placebo injection of saline (0.9% sodium chloride) as the comparator.Unlike steroid (triamcinolone), which is a white suspension, saline has the same appearance to adalimumab meant we were able to blind participants as to their treatment allocation.This was especially important given the subjective nature of some of the outcomes we planned to assess in the definitive trial, should this study have shown to be feasible.Interestingly, one of the main findings from our interviews was that participants understood the concept of randomisation and accepted the possibility of receiving placebo.
Despite not meeting the feasible objectives in terms of recruitment and success of the consent process, it was feasible to deliver the intervention in line with the trial protocol and all participants completed patient-reported and clinician assessed outcomes at 3 months.Data from the semistructured interviews showed that participants found the overall experience of the trial was positive.They participated in the AFF trial considering it as an opportunity to find a solution to their shoulder problem and also to benefit others.

Strengths
To our knowledge, AFF is the first study to assess the feasibility of conducting a large multicentre trial to test the effectiveness of a new treatment for people with early stage, pain-predominant frozen shoulder.This is important given the evidence for the limited effectiveness for treatments currently offered to people with frozen shoulder. 7 8 10 11 34he drug adalimumab has a very strong safety profile, having been used in over 5 million people, more than 25 000 trial participants and approved for nine different disorders. 35e used a targeted injection approach using ultrasound guidance, which was standardised across sites and meant we were able to maintain blinding of trial participants and members of trial team at sites.Despite the small number of participants recruited into the trial, our interviews provided a useful insight into patient experiences with the early painful phase of frozen shoulder and their perceptions about being involved in the AFF trial.AFF is one of the very few feasibility trials of upper limb conditions that include a qualitative component.

Limitations
A limitation of this study is that all participants were white British with an average age of 55 years.Therefore, the population recruited were not truly representative of the general population or those people affected by frozen shoulder, 36 despite recruiting from sites within inner cities with a more diverse population demographic.In addition, while all nine participants recruited into the trial received both their first and second injection within the specific timeframe, it is unclear whether this would be same for larger definitive trial with more pressure on treatment appointment slots.Similarly, while all participants completed both patient-reported and clinician-assessed outcomes at 3 months, it is unlikely this would be same for larger trial with longer follow-up.In a recent trial of corticosteroid injection and physiotherapy in people with a rotator cuff disorder, carried out across 20 NHS musculoskeletal services, the attrition rate at 6 and 12 months was 13%. 37It is also uncertain whether data saturation was reached from the eight semistructured interviews and, therefore.These findings may not be generalisable and should be interpreted only within the context of the AFF trial.

CONCLUSION
Our findings demonstrated that a definitive trial was not feasible within the NHS musculoskeletal and related physiotherapy services that participated in the AFF trial.This was largely due to difficulties in identifying patients while still in early-stage frozen shoulder and patient expectations regarding steroid injection.However, it was feasible to deliver the injection in line with trial protocol and to collect both patient and clinician assessed outcome data.Our interview findings indicated positive experiences from participants who took part in the trial.

Figure 1
Figure 1 Participant flow diagram.BESS British Elbow and Shoulder Society; TNF, tumour necrosis factor.