The clinical utility of ultrasonography for rotator cuff disease, shoulder impingement syndrome and subacromial bursitis

The lifetime prevalence of shoulder pain is identical to that of neck pain at 66%,1,2 ranking only behind the prevalence of low back pain (84%).3 Periarticular shoulder disorders are the most common cause of shoulder pain encountered by general practitioners and musculoskeletal specialists.4 Our decision to focus on rotator cuff disease (RCD), shoulder impingement syndrome (SIS) and subacromial bursitis (SAB) is based on the frequency with which these diagnoses appear in ultrasonography reports. Accurate clinical diagnosis of these disorders is often challenging.

Because RCD, SIS and SAB all have gradual or sudden onset of symptoms, affect the shoulder and/or lateral upper arm and cause painful or restricted shoulder movement, diagnosis based on the clinical history alone is unreliable. Diagnosis may be further confounded by the coexistence of different shoulder disorders, lack of consensus on diagnostic criteria, case definition and even nomenclature,5-8 and failure to establish what the patient means by “shoulder pain”. For example, patients frequently refer to the upper border of the trapezius and the scapula as their “shoulder”;7 pain here has different diagnoses from pain localised to the shoulder or lateral upper arm. If the clinician overlooks this and proceeds to shoulder ultrasonography, which has more than 50% chance of finding an abnormality in an asymptomatic shoulder,9,10 there is a risk of a type 2 diagnostic error (ie, diagnosis of a condition which either does not exist or which is not the cause of the patient’s symptoms).

Traditional clinical teaching, such as that pioneered by the English orthopaedic surgeon James Cyriax, focused on eliciting pain on active, passive and resisted shoulder movements to determine the site of the shoulder lesion.11,12 Such differentiation has been considered important in facilitating effective treatment.12,13 Although Cyriax described subacromial impingement of the rotator cuff, he never conceptualised this as a distinct syndrome with specific clinical tests. For detecting SIS, the clinician has an array of eponymous clinical tests designed to maximise or expose encroachment on the rotator cuff and subacromial bursa.8,14,15 The reliability of these tests depends on their sensitivity (ie, ability to detect the disorder when present) and specificity (ie, ability to avoid or minimise misdiagnosis when the disorder is absent). A systematic review that examined diagnostic clinical tests was unable to establish conclusive evidence of reliability16 so that, unsurprisingly, clinical studies have shown poor to moderate agreement and reproducibility between examiners.5,17-19 One study which did find good agreement, between two experienced physiotherapists,20 has been criticised because of the small number of cases and the paucity of medical information.5,19

The poor to moderate reliability and reproducibility of clinical tests for periarticular shoulder disorders5,16-19 may explain why Australian clinicians have turned increasingly to shoulder ultrasonography21 for diagnostic help. At issue is whether the reliability and reproducibility of shoulder ultrasonography makes it a more dependable diagnostic option than clinical examination. The appeal of ultrasonography lies in its safety, speed of operation, wide availability, relatively low cost and non-invasiveness.

Between 2000 and 2006, use of diagnostic ultrasound and ultrasound-guided shoulder injections increased in Australia (Box).21 The Medicare item number for ultrasound-guided shoulder injections includes other regions of the body, so its exact cost is unknown. The Medicare reimbursement for this procedure is $129.95, whereas for a blind intrasynovial injection (item number 50124) it is $22.25.21 A 2004 review found no evidence that imaging-guided subacromial bursal injections improved long-term outcomes,22 while a randomised study found that blind injections into the subacromial bursa by an experienced orthopaedic surgeon or by an experienced musculoskeletal radiologist were as accurate as ultrasound-guided injections.23

Reproducibility of a diagnostic test implies agreement between different examiners (ie, interexaminer agreement) and consistency of findings when the test is repeated by the same examiner (intra-examiner agreement). Reproducibility depends on the reliability (sensitivity and specificity) of the diagnostic test, the experience of the examiner37,45 and consensus on diagnostic criteria and nomenclature.6,7,45,46

In 2001, the EULAR (European League Against Rheumatism) Working Group for Musculoskeletal Ultrasound published technical guidelines in an effort to standardise scanning methods.45 This might have been expected to reduce interscanner differences in the acquired image, thus reducing interpretive differences. Yet, 4 years later, a prospective study by radiologists remained critical of the lack of standardisation of certain sonography procedures and noted “a lack of consensus in the radiology literature as to what appearances are normal and abnormal”.37 In 2006, investigators from the EULAR Working Group, in a study of interobserver reliability, found difficulty in reaching a consensus on “bursitis versus normality” and reported that enlarged bursae were “commonly detected with high resolution ultrasonography machines in normal subjects”.38 They considered that the paucity of studies on the “validity, reliability and sensitivity to change” had largely contributed to the perception of ultrasonography being the most operator-dependent imaging technique.38 It is sobering that despite ultrasonography being in clinical use for more than 30 years, standard reference values were published for the first time only in 2004.36

Because of the high incidence of abnormalities detectable by shoulder ultrasonography in asymptomatic patients,9,10,29,31,38 this technique is best employed for patients complaining of shoulder or lateral upper arm pain, particularly if pain is localised to these regions on clinical examination. This should reduce the risk of an incorrect diagnosis of a periarticular shoulder disorder in patients with nonspecific shoulder region pain, which is often localised to the trapezius or scapular regions and which is six times more common than rotator cuff disorder.47 Although neither the clinical examination nor ultrasonography is a perfect diagnostic instrument for shoulder pain, this is no reason to eschew either, but nor should there be an expectation that the sum of two imperfect instruments enhances diagnostic accuracy.18 Indeed, the high number of false positives in shoulder ultrasonography in asymptomatic patients9,10,29,31,38 suggests the opposite may be true.

Scepticism about the utility of clinical examination is not entirely justified. Studies have shown at least moderate inter-examiner agreement for RCD and SIS,15,19 the most common causes of shoulder pain. A systematic review suggested that clinical examination by specialists could rule out the presence of a rotator cuff tear,16 and both clinical and ultrasound studies suggest that inter-examiner reproducibility can be improved with training.5,19,38,45,46,48

On the available evidence, shoulder ultrasonography is best viewed as an adjunctive imaging modality in the diagnosis of shoulder pain rather than a stand-alone investigation. It has been suggested that shoulder ultrasound might be at its most valuable where the clinician is able to interpret the images in the setting of the clinical history and physical examination, “enabling ultrasound to become the physician’s extended finger”.45 In the real world this ideal situation is unlikely to eventuate unless the clinician is also a proficient ultrasonographer. In a government-subsidised, fee-for-service medical system such as operates in Australia, the issue of propriety might arise if the requestor of a service became its financial beneficiary.

Clinical diagnosis might be enhanced if there were agreement on ultrasound diagnostic criteria for shoulder disorders and better standardisation of radiology reports. In view of the Australian MBS exclusion of SIS from clinical indicators attracting a benefit in shoulder ultrasonography, it seems reasonable for radiologists to provide comment and clarification when the diagnosis of SIS is made, to explain the implications of a diagnosis of SAB, and in view of the many ultrasound-detectable abnormalities in asymptomatic individuals,9,10,29,31,38 to emphasise to clinicians the need for clinical correlation, whatever diagnosis is provided.

Shoulder ultrasonography is reliable in the diagnosis of full thickness tears but less reliable in the detection of partial thickness tears.16,29,30 The reliability of shoulder ultrasound in the diagnosis of SIS has not been clearly established. The validity of the diagnosis of SAB is problematic because of the lack of a suitable ultrasound definition of SAB,37 and because the “diagnosis” may not reflect a specific entity requiring specific treatment.44

Clinicians need diagnostic certainty to optimise management. A reported ultrasound abnormality enables the clinician to fulfil the doctor’s half of the bargain in the doctor–patient relationship, by delivering a diagnosis with the promise of treatment. The patient can be mollified and may even be satisfied. If the patient is also a claimant or litigant, the injury claim can seemingly be validated by an objectively verifiable abnormality on imaging. That all of this has the potential to reinforce and thereby perpetuate referral behaviour is consistent with the rise in requests for shoulder ultrasonography in Australia between 2000 and 2006.21 Yet on the available evidence, shoulder ultrasonography, like all other imaging modalities, is rarely able to deliver the diagnostic certainty for which clinicians yearn.