Aim and overview of trial findings
For the primary outcome, there was no evidence of a difference in the OSIS at the primary 6-month time point between participants randomised to receive advice only and those who received one session of advice and the offer of additional physiotherapy. At all time points the direction of change favoured a programme of physiotherapy; however, the 95% CI at each time point excluded the target difference of 4 points. There were no statistically significant differences in the QuickDASH or consistent differences in the EQ-5D-5L or complication secondary outcomes. Secondary unadjusted and per-protocol analyses, and a sensitivity analysis accounting for missingness, were not materially different.
Predefined subgroup analyses were undertaken to assess whether there was evidence that the intervention effect differed between age group (≤ 39 years old and ≥ 40 years old) or arm dominance. Our predefined subgroup analyses showed no evidence of clinically relevant effects from either age or arm dominance.
The probability of being cost-effective at a willingness-to-pay threshold of £30,000 for advice and the offer of additional physiotherapy was 0.946. Sensitivity analyses, including complete cases only and including broader societal costs, provided similar findings.
Qualitative experiences of involvement in ARTISAN were generally positive across both arms. There was a trend towards a more positive outcome reported by those in the ARTISAN Plus arm of the trial. There were some areas where there were differences related to age and participants’ requirements to return to sporting activities.
Generalisability
The ARTISAN trial was a pragmatic, multicentre RCT. People presenting at 41 UK hospitals in the NHS with a TASD for non-surgical management were randomised 1 : 1 to receive advice or advice and a programme of physiotherapy.
Two-thirds of participants were male, with the most common mechanism of injury being sport related. The mean age of participants was 45 years, and two age distributions centred at 24 and 56 years old, with the point where probability of membership in both distributions was 50% at 33 years of age. This was within 10 years of the randomisation stratum (age of 40) and therefore no further sensitivity analysis was required. These baseline demographics and two age distributions are reflective of the wider literature on first-time shoulder dislocation demographics.
Limitations
Among those screened and not meeting the eligibility criteria, the main reasons were that the patient was to undergo surgery (17%), the patient could not see a physiotherapist within 6 weeks (28%) or the patient could not adhere to trial procedures (39%).
A proportion of people with a first TASD having surgery is normal practice, and this trial was not addressing this surgical population. There was a higher than expected percentage of participants not being able to see a physiotherapist within 6 weeks, which on further discussion with sites was a result of added COVID pressures in some cases. The trial team did not collect further detailed information on the precise reasons why patients could not adhere to trial procedures. However, as this trial recruited before and during the pandemic there were changes in care pathways that needed to be accommodated in real time, such as a change to virtual clinics, which resulted in teams using this criterion in some cases.
The ARTISAN trial did have a loss to follow-up of 27%; however, the observed SD was much smaller than anticipated. This change in parameters reduced the number of participants required to observe the planned target difference of 4 points. The loss in participants was MAR, as it could be explained by our data. It was seen that the majority of the participants lost to follow-up were male (78%); however, this did not affect the balance between the intervention groups at the follow-up.
As the study was focused on the effectiveness of physiotherapy, there were concerns at the beginning of the study that there may be a physiotherapist effect which could strongly affect the estimate of efficacy of the ARTISAN intervention.
There were 96 therapists who delivered the core ARTISAN session across the 41 study sites. Only 42 therapists (44%) randomised more than three participants. An interim analysis demonstrated that the sample size did not need to be inflated for therapist effects. The analysis was repeated using the 6-month follow-up data at the end of the study to check whether the interim analysis was correct, which was confirmed.
Physiotherapists were trained to deliver the advice session and trial-related procedures. In the advice-only group, two participants did not adhere to the protocol due to a clinician mandating additional physiotherapy, and subsequently crossed over to receive a programme of physiotherapy. In the advice-only arm, all participants received the intervention.
Of those receiving the ARTISAN-only intervention, 15% subsequently had extra physiotherapy, whereas 83% of participants in the group allocated to additional physiotherapy received the additional physio sessions, demonstrating good levels of adherence.
Equality, diversity and inclusion
In preparation for this trial, patient and public groups were consulted who had experience of sustaining and being treated for a shoulder dislocation. These groups initially focused on local groups to the host organisation. Once an initial draft of trial materials and patient-facing documents were produced, these were presented to groups from a wider geographical area.
One of the areas of focus specific to the population who sustain a shoulder dislocation was the age range. Recognising that there are two dominant groups (young males who injure during sports and older women who injure during a fall), it felt inappropriate to produce materials that featured either young males or older females. Consequently, an approach using animations was chosen in an attempt to make the materials to apply to everyone equally.
In designing the trial materials, we ensured that plain simple language was used, there was a mix of media to access the same content (e.g. diagrams, videos, websites, booklets and text), and access to the materials was available regardless of access to electronic devices, to not exclude those who do not have regular access.
The qualitative study found that the use of animations was well received; however, some participants did comment on the difficulty of relating to some of the content because it was not three-dimensional or it was not clear exactly what was being moved and how to achieve the target exercise.
On reflection, although steps were taken to address the main area of ensuring that both key demographic groups felt the materials were appropriate, further work may be required to refine the animations or use a mixed medium of animations and real-life images to exemplify the rehabilitation content.
The trial population were broadly representative of the ethnic diversity of the UK. The trial materials were only available in the English language, so this is also an area for further development. Intervention materials will be made freely available to those who want to translate them as part of implementation of the findings.
Patient and public involvement
Patient and public involvement was at the heart of the ARTISAN study. Prior to the study, clinical co-applicants consulted with patients during appointments to ascertain whether the research gaps highlighted in the literature were of high importance to them. These same people were asked whether they would be interested in a consultation role for further development of the study and future roles in management of the research and dissemination of findings.
Subsequently, two patient representatives agreed, and discussed experiences and expectations of services and the plans for the trial. These perspectives were key in the development of the protocol to ensure trial processes, materials and interventions were feasible and acceptable. Key inputs from this group into the protocol were to ensure that the intervention employed a holistic approach, not just focusing on physical well-being.
One patient representative was included as a lay co-applicant who, in addition to contributing during our development work, was a member of the TMG. They contributed to trial processes and paperwork, such as patient information leaflets. They took a lead in the development of the information, training materials and resources used within the study. As the trial progressed, in the later stages our PPI team members changed, but continued in the same role with the same aims. Going forwards, they will be key to us ensuring that we disseminate our findings to a wider audience. A second PPI member agreed to an oversight role on the independent TSC for the duration of the trial and advised on the final report and dissemination plans.
Overall evidence
ARTISAN is the largest RCT that has investigated two different two rehabilitation interventions in adults with a first-time traumatic shoulder dislocation. We did not find any statistically significant differences in the primary outcome (OSIS) or other secondary outcomes. Advice and a programme of physiotherapy was cost-effective at a £30,000/QALY willingness-to-pay threshold.
While the economic interpretation is supportive of ARTISAN plus additional physiotherapy, the clinical interpretation may be more nuanced: where health gains appear small and uncertain and physiotherapy resources are fully stretched meeting current service demands, a programme of physiotherapy may not be perceived by practitioners as a service priority.
A 2014 Cochrane Review did not find any RCTs that compared different rehabilitation methods after the initial 2 weeks of supporting the arm in a sling. An updated review in 2019 had the same conclusion, but identified one ongoing study that has since completed.8,9 No further ongoing studies were identified in an updated search of trial registries.
The now completed single RCT randomised 56 participants, across three orthopaedic shoulder units in Denmark, to a home-based exercise intervention or a supervised 12-week intervention. This trial included participants under the age of 40 years only and compared two different rehabilitation approaches (neuromuscular exercises vs. strengthening exercises), with one being delivered in a physiotherapy clinic and the other being home-based. Additionally, the physiotherapy-based programme was a prescriptive 14 sessions of 45 minutes each. ARTISAN included all adults and compared the same rehabilitation approach, with the only difference being delivery by a physiotherapist versus delivery at home; the ARTISAN protocol also allowed a tailored approach determined by the physiotherapist, rather than being prescriptive. This heterogeneity means it is not appropriate to consider pooling results from the two studies.
The authors reported a statistically significant improvement in the primary outcome (Western Ontario Shoulder Instability Index) from clinic-based neuromuscular exercise when compared to home exercises. The point estimate was less than their predefined clinically relevant between-group difference of 250, although a clinically relevant difference was not excluded (between-group mean difference, –228.1, 95% CI –430.5 to –25.6; p = 0.028). In ARTISAN, although the point estimate favoured physiotherapy, this was not statistically significant, and the limits of the 95% CI excluded our target difference. Overall, these findings, taken in conjunction with the ARTISAN health economic analysis, mean the possibility that there is a small mean benefit from additional physiotherapy after a first shoulder dislocation has not been totally excluded. However, any such benefit is likely to be very small and not of clinical importance or relevance for most people with a shoulder dislocation.