Surgical treatments compared with early structured physiotherapy in secondary care for adults with primary frozen shoulder: the UK FROST three-arm RCT

Background Frozen shoulder causes pain and stiffness. It affects around 10% of people in their fifties and is slightly more common in women. Costly and invasive surgical interventions are used, without high-quality evidence that these are effective. Objectives To compare the clinical effectiveness and cost-effectiveness of three treatments in secondary care for adults with frozen shoulder; to qualitatively explore the acceptability of these treatments to patients and health-care professionals; and to update a systematic review to explore the trial findings in the context of existing evidence for the three treatments. Design This was a pragmatic, parallel-group, multicentre, open-label, three-arm, randomised superiority trial with unequal allocation (2 : 2 : 1). An economic evaluation and a nested qualitative study were also carried out. Setting The orthopaedic departments of 35 hospitals across the UK were recruited from April 2015, with final follow-up in December 2018. Participants Participants were adults (aged ≥ 18 years) with unilateral frozen shoulder, characterised by restriction of passive external rotation in the affected shoulder to < 50% of the opposite shoulder, and with plain radiographs excluding other pathology. Interventions The inventions were early structured physiotherapy with a steroid injection, manipulation under anaesthesia with a steroid injection and arthroscopic capsular release followed by manipulation. Both of the surgical interventions were followed with post-procedural physiotherapy. Main outcome measures The primary outcome and end point was the Oxford Shoulder Score at 12 months post randomisation. A difference of 5 points between early structured physiotherapy and manipulation under anaesthesia or arthroscopic capsular release or of 4 points between manipulation under anaesthesia and arthroscopic capsular release was judged clinically important. Results The mean age of the 503 participants was 54 years; 319 were female (63%) and 150 had diabetes (30%). The primary analyses comprised 473 participants (94%). At the primary end point of 12 months, participants randomised to arthroscopic capsular release had, on average, a statistically significantly higher (better) Oxford Shoulder Score than those randomised to manipulation under anaesthesia (2.01 points, 95% confidence interval 0.10 to 3.91 points; p = 0.04) or early structured physiotherapy (3.06 points, 95% confidence interval 0.71 to 5.41 points; p = 0.01). Manipulation under anaesthesia did not result in statistically significantly better Oxford Shoulder Score than early structured physiotherapy (1.05 points, 95% confidence interval –1.28 to 3.39 points; p = 0.38). No differences were deemed of clinical importance. Serious adverse events were rare but occurred in participants randomised to surgery (arthroscopic capsular release, n = 8; manipulation under anaesthesia, n = 2). There was, however, one serious adverse event in a participant who received non-trial physiotherapy. The base-case economic analysis showed that manipulation under anaesthesia was more expensive than early structured physiotherapy, with slightly better utilities. The incremental cost-effectiveness ratio for manipulation under anaesthesia was £6984 per additional quality-adjusted life-year, and this intervention was probably 86% cost-effective at the threshold of £20,000 per quality-adjusted life-year. Arthroscopic capsular release was more costly than early structured physiotherapy and manipulation under anaesthesia, with no statistically significant benefit in utilities. Participants in the qualitative study wanted early medical help and a quicker pathway to resolve their shoulder problem. Nine studies were identified from the updated systematic review, including UK FROST, of which only two could be pooled, and found that arthroscopic capsular release was more effective than physiotherapy in the long-term shoulder functioning of patients, but not to the clinically important magnitude used in UK FROST. Limitations Implementing physiotherapy to the trial standard in clinical practice might prove challenging but could avoid theatre use and post-procedural physiotherapy. There are potential confounding effects of waiting times in the trial. Conclusions None of the three interventions was clearly superior. Early structured physiotherapy with a steroid injection is an accessible and low-cost option. Manipulation under anaesthesia is the most cost-effective option. Arthroscopic capsular release carries higher risks and higher costs. Future work Evaluation in a randomised controlled trial is recommended to address the increasing popularity of hydrodilatation despite the paucity of high-quality evidence. Trial registration Current Controlled Trials ISRCTN48804508. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 24, No. 71. See the NIHR Journals Library website for further project information.