Study reference

Study characteristics

Patient characteristics

Index test

(test of interest)

Reference test

Outcome measures and effect   size

Level of evidence

Comments / limitations of   study

Diagnostic test accuracy

Hegedus, 2008

Type of study:

Systematic review and meta-analysis (diagnostic test accuracy)

Source of funding:

No competing interests

Inclusion criteria for studies:

- criterion standard was

surgery, magnetic resonance imaging (MRI) or injection (acromioclavicular   joint only)

- at least one physical examination test/special test was studied

- one of the paired statistics of sensitivity and specificity were   reported or could be discerned for an individual test,

- the article was in the English language

Exclusion criteria for studies:

- the special test was performed under anaesthesia or in cadavers

- a group of special tests was assigned the status of ‘‘composite   physical examination’’

- the article was a review

Search until October 2006

N= 21 studies on impingement or rotator cuff integrity

Mean age ± SD:

Not reported

Sex: not reported

- Impingement tests

- Rotator cuff (RC) integrity tests

- Glenoid labrum integrity tests

- Acromioclavicular (AC) joint tests

- Surgery

- MRI

 - Injection (AC joint only)

Sensitivity (SE), specificity (SP), 95% confidence interval (95%CI)

Impingement

Neer test (4 studies, n=834 shoulders, pooled):

SE: 79% (95%CI 75-82)

SP: 53% (95%CI 48-58)

Hawkins-Kennedy test (4 studies, n=834 shoulders, pooled):

SE: 79% (95%CI 75-82)

SP: 59% (95%CI 53-64)

Rotator cuff integrity

Supraspinatus/Jobe/empty can test (7 studies, n=1356 shoulders,   range):

SE: 41-89%

SP 50-98% (6 studies)

A1: systematic   review of at least 2 independent   studies of level A2 (index test compared to valid reference standard;   cut-offs were defined a priori; independent interpretation of test results;   an adequate number of patients were enrolled consecutively; all patients   received both tests)

- Study quality assessed with Quality Assessment of Diagnostic   Accuracy Studies (QUADAS) tool (14 items, max. score 14)

- Higher quality study defined as score ≥ 10

- 2/6 studies on impingement were of higher quality

- 8/15 studies on rotator cuff integrity were of higher quality

- Meta-analysis for the Neer test of impingement, the Hawkins-Kennedy   test of impingement and the Speed test for a SLAP lesion shows these tests to   have no discriminatory ability for shoulder diagnosis.

- Meta-analysis for other tests was not possible either due to lack of   sufficient studies or heterogeneity between studies.

- Review included studies that reported only sensitivity or   specificity

- Typographical error: specificity of 98% reported for empty can test,   not appearing to reflect the value of 50% reported in the original study   (Itoi et al 1999).

Hughes,

2008

Type of study:

Systematic review (diagnostic test accuracy)

Source of funding:

Not reported

Inclusion criteria for studies:

- Studies in peer-reviewed journals

- English language studies

- Human participants

- Subjects presenting with shoulder problems

- Clinical diagnostic testing for rotator cuff pathology (tear or   inflammatory change)

- Clinical tests used are primarily for the diagnosis of rotator cuff   pathology

- Results of the clinical tests are compared to the findings of a   reference standard – MRI or operation report

- Sufficient data are presented to allow calculation of specificity   and sensitivity for the clinical tests

Search until April 2007

N=13 studies with 42-552 participants

Age (range): 24 to over 77 years

Sex: 56% M / 44% F (data from 10 studies)

- Impingement tests

- Supraspinatus tests

- Infraspinatus tests

- Subscapularis tests

- MRI or operation report

3 additional studies compared to review of Hegedus, 2008. Effects   comparable to effects found in review of Hegedus, 2008.

B: systematic review of   lesser-quality studies (only 1 A2-level study included in the review)

- Study quality assessed with 5 criteria adapted from guidelines for   appraising studies concerned with diagnostic tests by the National Health and   Medical Research Council (1999).

- Higher quality study not defined.

- Meta-analysis was not performed due to the variety of methods and   tests used across the studies.

- Hegedus et al (2008) examined 10 of the 13 papers included in this   review and, overall, the poor accuracy of clinical tests for rotator cuff   pathology demonstrated in this review was found in the results of Hegedus et   al. (2008).

Author’s conclusions:

- Overall, most tests for rotator cuff pathology were inaccurate and   cannot be recommended for clinical use.

- At best, suspicion of a rotator cuff tear may be heightened by a   positive palpation, combined Hawkins/painful arc/infraspinatus test, Napoleon   test, lift-off test, belly-press test, or drop-arm test, and it may be   reduced by a negative palpation, empty can test or Hawkins-Kennedy test.

Michener, 2009

Type of study:

Diagnostic test accuracy study

Source of funding:

No commercial party having a direct financial interest in the results   of the research

supporting this article has or will confer a benefit on the authors or   on any organization

with which the authors are associated.

Inclusion criteria:

- Consecutive patients presenting with shoulder pain to an orthopedic   surgeon’s office

- patients had to report shoulder pain for at least 1 week, and   shoulder pain had to be their primary complaint.

Exclusion criteria:

Not reported

N=55 patients

Age,mean ± SD (range): 40.6 ± 15.1 years (range 18–83y)

Sex: 47 M/8 F

5 shoulder tests:

Neer, Hawkins-Kennedy, painful arc, empty can (Jobe), and external   rotation resistance test

arthroscopy

ROC curve analysis:

The cut point to discriminate between patients with and without SAIS was 3 positive   tests out of 5 (AUC=0.79, 95% CI 0.66–0.92; P=0.001).

Diagnostic accuracy for any test combination (≥ 3 out of 5 positive):

SE: 0.75 95% CI 0.54-0.96)

SP: 0.74 (95% CI 0.61-0.88)

LR+: 2.93 (95% CI 1.60-5.36)

LR-: 0.34 (95% CI 0.14-0.80)

Posttest probabilities:

≥ 3 out of 5 positive: LR+= 54.4%

< 3 out of 5 positive: LR- = 12.1%

B: Index test compared to reference test, but without all the features   mentioned for level A2.

Small sample size (n=55), unclear if the index test results were   interpreted without knowledge of the results of the reference standard.

Author’s conclusion:

The single tests of painful arc, external rotation resistance test,   and empty

can provide the best diagnostic utility and reliability. The Neer test   has clinical utility to screen for SAIS   but has only fair

reliability. Also of diagnostic utility is the use of the cut point of   3+/5 tests, with 3 or more tests positive of 5 useful in confirming SAIS,   whereas less than 3 positive of the 5 tests is helpful in decreasing the   likelihood of SAIS.

Murrell, 2001

Type of study:

Diagnostic test accuracy study

Source of funding:

St George Hospital/South Eastern Sydney Area Health Service, St George

Private Hospital/Health Care of Australia, and Mitek Australia   Ltd

Inclusion criteria:

All patients were referred to the practice of an orthopaedic surgeon   at St George Private Hospital, Sydney,    Australia,   between May, 1996, and December, 1999, and had shoulder injuries sufficient   to

warrant arthroscopic examination.

Exclusion criteria:

Not reported

N=400 patients (n=200 patients with rotator cuff tear and n=200   patients with no RC tear)

Age (range): 20-79 years

Sex: not reported

23 commonly used shoulder tests wasting, tenderness, and signs (the   droparm sign, O’Brien’s sign, and impingement), as well as active and passive   ranges of motion, and shoulder strength.

arthroscopy

Study 1 (n=100 with RC tear and n=100 with no tear):

tests for three clinical features (supraspinatus weakness; weakness in   external rotation; and impingement (in internal rotation, in external   rotation, or in both directions)) were more positive in patients with rotator   cuff tears than in the no tear set (p<0.0001).

study 2 (n=200 with RC tear and n=200 with no tear, verification of   findings from study 1):

(age-adjusted) post-test probability:

0.98 (95%CI 0.89-1.00) chance of RC tear if all three clinical   features test positive for any age, or if 2 out of 3 tests are positive and   patients is aged ≥ 60 years, or if drop-arm sign test is positive for any   patient.

0.05 (95% CI 0.02-0.11) chance of RC if none of the three clinical   features test positive for any patient.

B: Index test compared to reference test, but without all the features   mentioned for level A2.

Unclear if patients were enrolled consecutively and if the index test   results were interpreted without

knowledge of the results of the reference standard.

Author’s conclusion:

The predictive power of the combined clinical tests is

similar to the best values for magnetic resonance imaging and   ultrasonography. Moreover, the clinical tests are easy and convenient to do.

Park, 2005

Type of study:

Diagnostic test accuracy study

Source of funding:

The authors did not receive grants or outside funding in support

of their research or preparation of this manuscript. They did not

receive payments or other benefits or a commitment or agreement

to provide such benefits from a commercial entity

Inclusion criteria:

patients who had shoulder surgery performed by the senior author

between August 1992 and June 2003.

Exclusion criteria:

patients who did not undergo arthroscopy were excluded

from the study.

N=552 patients (n=193 nonimpingement and n=359 impingement)

From 1127 eligible patients, 214 did not undergo arthroscopy, 361 were   excluded for the following reasons: a history of shoulder surgery (n=126); AC   arthritis, requiring excision of the distal part of the clavicle in   conjunction with acromioplasty or rotator cuff repair, combined with   impingement syndrome (n=73); an incomplete physical examination because of   limited motion or extreme pain (n=63); superior labrum

anterior-to-posterior lesions, requiring repair, combined with

impingement syndrome (n=57); or instability combined with impingement   syndrome (n=42).

Age and sex not reported.

8 clinical tests

for impingement syndrome:

Neer test

Hawkins-Kennedy test

Painful arc sign

supraspinatus muscle strength test performed according to the   technique described by Jobe and Moynes

Speed test

cross-body adduction test

drop-arm sign described by Codman

infraspinatus muscle strength test (external rotation)

arthroscopy

Multiple logistic regression analysis

Bursitis:

Neer test (p<0.0001)

Partial-thickness RC tear:

Neer test (p=0.007)

Full-thickness RC tear:

Painful arc sign (p<0.0001)

Drop-arm sign (p=0.024)

Infraspinatus test (p=0.06)

Overall impingement syndrome:

Hawkins-Kennedy test (p<0.0001)

Painful arc sign (p<0.0001)

Infraspinatus test(p<0.0001)

Posttest probabilities for combination of tests

Overall impingement syndrome (bursitis, PT RC tear and/or FT RC tear)   (n=352 patients):

0.95 chance of any degree of impingement syndrome if all three tests   are positive (Hawkins-Kennedy impingement sign, painful arc sign, and   infraspinatus muscle test)

0.24 chance of any degree of impingement syndrome if none of the three   tests are positive.

Full-thickness rotator cuff tears (n=348 patients):

0.91 chance of FT RC tear if all three tests are positive (painful arc   sign, drop-arm sign, and infraspinatus muscle test)

0.09 chance of FT RC tear if if none of the three tests are positive.

B: Index test compared to reference test, but without all the features   mentioned for level A2.

Unclear if patients were enrolled consecutively and if the index test   results were interpreted without

knowledge of the results of the reference standard, no 95% CIs   reported.

Author’s conclusion:

The severity of the impingement syndrome affects the diagnostic values   of the commonly used clinical

tests. The variable accuracy of these tests should be taken into   consideration when evaluating patients with symptoms of rotator cuff disease.

Reliability

May, 2010

Type of study:

Systematic review (reliability)

Source of funding:

No conflicts of interests

Inclusion criteria for studies:

- Results published as full reports before June 2009.

- The study involved physical examination   procedures used in shoulder examination.

- The study involved human subjects with any   sort of shoulder pain.

- The study had to be an intra- and/or   inter-examiner reliability design.

- The study had to be available in English.

- The study did not involve a mechanical device,   but simple tape measures were accepted as commonly available.

- The study did not involve subjects with   non-musculoskeletal conditions.

- The study did not involve asymptomatic   volunteers alone, although studies including a mix of symptomatic and

asymptomatic participants were included.

Search until June 2009

N=36 studies (number or participants not reported)

Mean age ± SD: not reported

Sex: not reported

- Impingement/rotator cuff tests

- AC joint tests

- glenoid labral pathology tears tests

Not applicable

Inter-tester reliability (kappa)

Neer test (4 studies, range):

010-1.00

Hawkins-Kennedy test (4 studies, range):

0.18-0.91

Supraspinatus/Jobe/empty can test (4 studies, range):

0.44-0.94

Intra-tester reliability (kappa)

Neer test (1 study, n=33):

1.00

Hawkins-Kennedy test (1 study, n=33): 1.00

Supraspinatus/Jobe test (1 study, n=33):

1.00

A1: systematic   review of at least 2 independent   studies of level A2 (high quality study)

- study quality assessed with a checklist (14 items, max. score 100)

- Higher quality study defined as score > 60%

- 17/36 (47%) studies had score > 60%

- review probably A1-level, but there are no established or commonly   used criteria for judging the quality of reliability studies.

- A meta-analysis was not attempted due to the heterogeneity of tests,   patients and analyses, and as direct comparison of reliability studies was   deemed inappropriate.

- Predetermined criteria for satisfactory reliability was 0.85.

- According to the study criteria, none of the tests demonstrated   consistent and acceptable levels of reliability, with overall evidence being   contradictory.

- Approximately half of the studies were of high quality, with   high-quality studies less likely to meet the criteria for acceptable   reliability.