Statistics from Altmetric.com
Hay and colleagues concluded, in their extended report of a trial of physiotherapy and injection for unilateral shoulder pain that physiotherapy and local steroid injection are of similar effectiveness.1 They suggest that “The high overall success rates... argue against the need for further exploratory trials in this condition”. I disagree. A large number of studies of shoulder pain have been bedevilled by diagnostic criteria that are not precise,2 and this study must unfortunately join the others.
Unilateral shoulder pain has a number of different causes. The study by Hay excludes a few specific conditions—in particular, a ruptured rotator cuff, but must by definition include a heterogeneous group of problems that are in fact quite discrete. These include frozen shoulder (adhesive capsulitis), rotator cuff injuries without full rupture, subacromial joint arthritis (sometimes known as subacromial bursitis), bicipital tendinitis, acromioclavicular joint disease, and subdeltoid bursitis. It is barely credible to imagine that several of these could be successfully treated by a steroid injection into the subacromial joint. In particular, the subacromial joint does not communicate with the glenohumeral joint unless the rotator cuff is ruptured, so frozen shoulder cannot be treated with a subacromial injection. Thus any study of shoulder pain must separate the different causes into different groups. Others have done this and shown that the relative benefits of physiotherapy and injection may be different.3
Furthermore, it must be clear that any clinicians contributing to a trial are working to the same diagnostic criteria. Even experienced consultant rheumatologists cannot agree on exact diagnoses, as I and colleagues have shown previously,4 and we concluded “...recruitment of patients for studies of the treatment of shoulder lesions requires care to avoid selection of a heterogeneous group”. Given the variability of rheumatology training and experience in general practice it seems unlikely that diagnostic precision will be sufficient in that setting, and Hay’s study does not conform to our recommendation.
For these reasons, far from suggesting that no further research is needed, this study underlines the need for clear and exact diagnostic criteria and further treatment trials for each of the specific causes of unilateral shoulder pain. Partly because clinical diagnosis may be difficult the use of magnetic resonance imaging scanning to define pathology may be an essential part of the investigation before treatment; I have certainly encountered many patients where a clear clinical picture is belied by a scan, particularly in the identification of rotator cuff pathology.
My own local audit of about 800 referrals of patients with shoulder pain suggested that some 40% had a frozen shoulder, with another 40% having abnormality in the rotator cuff/subacromial joint mechanism. Thus one could argue that a pair of steroid injections, one into the glenohumeral joint and one into the subacromial joint, might be expected to benefit about 80% of patients. However, such an approach, while practical, will not resolve in a scientific way the continuing uncertainty over the management of shoulder pain.
We agree with Bamji that, although a number of ways of classifying shoulder problems have been proposed, none have been shown to be valid or reliable. In our large primary care trial1 we adopted the “red flag” approach and are clear about its limitations in the paper. Although our trial was pragmatic, it examined a clearly stated question—“in patients presenting to general practitioners with a new episode of unilateral shoulder pain, and in whom specific “red flag” problems have been excluded, is a subacromial injection or a course of physiotherapy the best first choice?”.
Essentially, we were comparing two treatments commonly used by general practitioners. We were not investigating the relative merits of different types of injection, or indeed which components of the physiotherapy package had specific benefits. These are different questions, important in their own right, but not the ones we chose to answer in this particular study.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.