Hawkins–Kennedy test

Written by manual therapist Roar Syltebø

The Hawkins–Kennedy test was first described by Hawkins and Kennedy in 1980 (Hawkins & Kennedy, 1980). The test is designed to provoke the subacromial structures against the acromion and the coracoacromial ligament. In several studies the test has shown high sensitivity for subacromial impingement, but highly variable specificity (Calis et al., 2000; Hegedus et al., 2012). Used together with other tests, such as the painful arc and resisted external rotation, a test cluster can give higher specificity (Park et al., 2005; Michener et al., 2009). The test is of little use for establishing a specific tissue diagnosis, but can be useful when pain from the subacromial space and the rotator cuff is suspected.

Indication

The Hawkins–Kennedy test is indicated in pain in the shoulder region, with or without referral to the shoulder girdle or the arm. The pain may vary from diffuse to sharp and shooting, possibly with a sensation of pain radiating down the arm.

Procedure

The test is performed with the patient sitting. The examiner stands beside or behind the patient. The patient’s scapula is stabilised, and the examiner brings the arm into 90 degrees of flexion with the elbow flexed to 90 degrees. The shoulder is then passively moved into full internal rotation, and the examiner applies gentle overpressure into internal rotation. The test can be performed with the arm in varying degrees of horizontal adduction and abduction.

Interpretation

The Hawkins–Kennedy test is a pain provocation test. The test is positive when it reproduces the patient’s familiar pain in the subacromial and anterolateral region of the shoulder. The manoeuvre itself is uncomfortable for most people. Discomfort and pain elsewhere, or pain the patient does not recognise, do not constitute a positive test.

In a cadaver study, Valadie and colleagues (2000) examined what happens anatomically during the test. In the Hawkins position, the soft tissue came into contact with the coracoacromial ligament in all four shoulders examined. At the same time, the articular side of the rotator cuff tendons came into contact with the anterosuperior rim of the glenoid. In one shoulder, the subscapularis tendon was indented by the coracoid process (Valadie et al., 2000). The test therefore provokes both the subacromial side and the articular side of the rotator cuff. This may explain some of its low specificity, since internal impingement can also produce a positive test.

Terminology has changed since the test was described. The 2014 Dutch Orthopaedic Association guideline for subacromial pain syndrome considers the term subacromial impingement syndrome outdated, because the idea of the rotator cuff being mechanically pinched does not fully explain the condition. The Dutch guideline therefore prefers the term subacromial pain syndrome (SAPS), and states that the diagnosis can only be made with a combination of clinical tests (Diercks et al., 2014). Salamh and Lewis (2020) go further, arguing that special tests for rotator cuff-related shoulder pain, such as the Hawkins–Kennedy test and Neer’s test, cannot identify which structure is causing the pain, and that they should be removed from the examination (Salamh & Lewis, 2020).

Reliability

Johansson and Ivarson (2009) examined the inter-examiner and intra-examiner reliability of Neer’s test, the Hawkins–Kennedy test, Jobe’s supraspinatus test and Patte’s manoeuvre. They found very good reproducibility for all the tests, with kappa between 0.91 and 1.0 (Johansson & Ivarson, 2009). In contrast, Michener and colleagues (2009) found only fair inter-examiner reliability for the Hawkins–Kennedy test and Neer’s test, with kappa of 0.39–0.40, whereas the painful arc, the empty can test and resisted external rotation showed moderate to good reliability (kappa 0.45–0.67) (Michener et al., 2009).

Diagnostic accuracy

Calis and colleagues (2000) examined 120 patients (125 shoulders) with shoulder pain. The reference standard for subacromial impingement syndrome was relief from a diagnostic block in the subacromial space: a patient was classified as a case of subacromial impingement syndrome if they experienced marked pain relief after injection into the subacromial space, and the remaining participants formed the control group. The Hawkins–Kennedy test had a sensitivity of 92.1% and a specificity of 25.0%. In addition, the patients were divided into three groups according to the stage of rotator cuff involvement on MRI, as defined by Zlatkin and colleagues (1989). Group 1: increased signal without irregularity or tear of the tendons. Group 2: increased signal with irregularity and partial-thickness tear. Group 3: full-thickness tear of the supraspinatus tendon (Calis et al., 2000; Zlatkin et al., 1989).

Group 1 Group 2 Group 3
Sensitivity (%) 95.2 87.5 100
Specificity (%) 30.7 23.0 35.7
Accuracy (%) 59.5 64.8 73.5
Positive predictive value (%) 52.6 67.7 68.9
Negative predictive value (%) 88.8 50.0 100.0
The Hawkins–Kennedy test in different stages of subacromial impingement syndrome (Calis et al., 2000)

Park and colleagues (2005) found a sensitivity of 71.5% and a specificity of 66.3% for subacromial impingement syndrome, with a positive predictive value of 79.7% and a negative predictive value of 55.7%. The patients were divided into three groups according to severity: rotator cuff involvement and bursitis without a tear (group 1), partial-thickness tear (group 2) and full-thickness tear (group 3). Sensitivity was 75.7%, 75.4% and 68.7%, and specificity 44.5%, 44.4% and 48.3%, respectively. The combination of the Hawkins–Kennedy test, the painful arc and resisted external rotation (infraspinatus test) gave the highest post-test probability (95%) of impingement of any grade (Park et al., 2005).

MacDonald and colleagues (2000) examined the sensitivity and specificity of Neer’s test and the Hawkins–Kennedy test for subacromial bursitis and rotator cuff pathology. The Hawkins–Kennedy test had a sensitivity of 92% and a specificity of 44% for subacromial bursitis, and a sensitivity of 88% and a specificity of 43% for rotator cuff pathology. Considered together, the two tests had a very high negative predictive value, 96% for subacromial bursitis and 90% for rotator cuff pathology (MacDonald et al., 2000).

Michener and colleagues (2009) found that the Hawkins–Kennedy test was the only one of five tests for subacromial impingement that did not discriminate significantly between patients with and without the diagnosis (area under the ROC curve). In the test cluster of the Hawkins–Kennedy test, Neer’s test, the painful arc, the empty can test and resisted external rotation, the probability of subacromial impingement syndrome increased when three or more tests were positive, and decreased when fewer than three were positive (Michener et al., 2009).

Most of the studies above were carried out in patients in secondary care, many of them with surgical findings as the reference standard. Cadogan and colleagues (2012) instead examined 196 patients from primary care, with an ultrasound-guided diagnostic block of the subacromial bursa as the reference standard. In all, 34% experienced at least 80% pain relief after the injection. The Hawkins–Kennedy test was positive in 58% of patients with a positive response and in 67% of those without, and thus did not distinguish patients with subacromial pain from others. On the contrary, a negative Hawkins–Kennedy test was weakly associated with a positive response, although the association was not statistically significant. The strongest predictors were a strain injury as the mechanism of onset, anterior shoulder pain, and passive external rotation at 90 degrees of abduction not provoking the pain. The authors point out that the anaesthetic may also have reached adjacent structures, such as the rotator cuff and the AC joint (Cadogan et al., 2012).

Cook and Hegedus (2008) also present the Hawkins–Kennedy test as a test for superior labral tears, but according to the studies they cite, it performs poorly for this purpose (Cook & Hegedus, 2008, p. 94). We therefore see no reason to use the Hawkins–Kennedy test for anything other than its intended purpose: a test for subacromial impingement.

Because the inferior aspect of the AC joint forms part of the roof of the subacromial space, the Hawkins–Kennedy test may in theory also provoke pain from the AC joint, for example in osteoarthritis with inferior osteophytes. However, the test cannot distinguish between subacromial pain and AC joint pain, and is therefore of little value when assessing the AC joint. A primary care study using a diagnostic block of the AC joint as the reference standard found that the Hawkins–Kennedy test did not change the probability of AC joint pain (Cadogan et al., 2013). Other tests and clinical findings are better suited for this purpose.

Reviews and meta-analyses

In an updated meta-analysis, Hegedus and colleagues (2012) found a pooled sensitivity of 79% and a specificity of 59% for the Hawkins–Kennedy test for subacromial impingement. According to the authors, the test may at first glance appear useful for ruling out impingement, but its negative likelihood ratio is too weak for this. None of the three impingement tests in the analysis (the Hawkins–Kennedy test, Neer’s test and the painful arc) produced likelihood ratios that meaningfully change the probability of the diagnosis (Hegedus et al., 2012).

Alqunaee and colleagues (2012) included ten studies with a total of 1,684 patients in a meta-analysis, all with surgery as the reference standard. The Hawkins–Kennedy test, Neer’s test and the empty can test had higher pooled sensitivity (0.69–0.78) than specificity (0.57–0.62). The authors conclude that the tests are more useful for ruling out than for ruling in subacromial impingement (Alqunaee et al., 2012).

Gismervik and colleagues (2017) found that the Hawkins–Kennedy test had the highest diagnostic odds ratio of the impingement tests, but it was only 2.86, with a sensitivity of 0.58 and a specificity of 0.67. No single test was superior (Gismervik et al., 2017). A 2013 Cochrane review included 33 studies of physical tests for impingement. The tests were performed and interpreted so differently that the results could not be pooled, and most of the studies were carried out in patients awaiting surgery. The authors found insufficient evidence to recommend specific tests in primary care (Hanchard et al., 2013).

Cook and Hegedus (2012) give the Hawkins–Kennedy test a utility score of 3, meaning that the evidence minimally supports or does not support its use (on a scale from 1, strong support, to 3, minimal or no support). This assessment reflects the literature up to 2012. More recent reviews confirm that the test is sensitive but not very specific, and that on its own it changes the probability of the diagnosis only slightly (Gismervik et al., 2017; Hanchard et al., 2013).

Body Examination suggests the following interpretation of the Hawkins–Kennedy test:

  • Pain provoked anterolaterally in the shoulder, which the patient recognises as their familiar symptoms, suggests subacromial impingement syndrome (subacromial pain syndrome).
  • The Hawkins–Kennedy test is sensitive but not very specific. The test provokes both the subacromial side and the articular side of the rotator cuff, and therefore cannot distinguish between the structures that may be causing the pain.
  • The test should be used for screening together with Neer’s test, active abduction (painful arc) and resisted isometric tests. Three or more positive tests increase the probability of subacromial pain syndrome.
  • Clinically, it is often less important to distinguish which structures are causing the shoulder pain, and the test is considered useful as part of a test cluster for determining whether the symptoms are likely to arise from the shoulder and the subacromial region. As always, the findings must be interpreted in light of the history and the preceding examination.
  • If a more tissue-specific diagnosis is important for further management, supplementary ultrasound examination is recommended, with the findings assessed against the clinical picture.

Literature search, updating and translation were carried out with the assistance of AI (Claude, Anthropic). All content has been reviewed and approved by Editor-in-Chief Roar Syltebø.

References

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.