Diagnosis of Shoulder Pain by History and Selective Tissue Tension: Agreement Between Assessors

Study Design: Evaluation of agreement between assessors. Objective: To evaluate agreement between an expert in selective tissue tension (STT) and 3 other trained assessors, all using STT in conjunction with a preliminary clinical history, on their diagnostic labeling of painful shoulders. Background: Consensus on diagnostic labeling for shoulder pain is poor, hampering interpretation of the evidence for interventions. STT, a systematic approach to physical examination and diagnosis, offers potential for standardization, but its reliability is contentious. Methods and Measures: Four trained assessors, 1 of whom was considered an expert, separately assessed 56 painful shoulders in 53 subjects (32 male [mean ± SD age, 51 ± 13 years], 21 female [mean ± SD age, 57 ± 12 years]), using STT in conjunction with a preliminary clinical history. Assessors labeled each painful shoulder as ‘‘rotator cuff lesion,’’ ‘‘bursitis,’’ ‘‘capsulitis,’’ ‘‘other diagnosis,’’ or ‘‘no diagnosis.’’ Combinations of diagnoses were allowed. Results: A diagnosis was made in every case, with less than 7% of the diagnoses being combined. With the diagnostic categories pooled, agreement (kappa and 95% confidence interval [CI]) between the expert assessor and each of the other assessors was good, ranging from 0.61 (0.44-0.78) to 0.75 (0.60-0.90). For single diagnostic categories, agreement between the expert and each of the others (dichotomized data) ranged from 0.35 (–0.03-0.73) to 0.58 (0.29-0.87) for bursitis; 0.63 (0.40-0.86) to 0.82 (0.65-0.99) for capsulitis; 0.71 (0.49-0.93) to 0.79 (0.61-0.96) for rotator cuff lesions; and from 0.69 (0.35-1.00) to 0.78 (0.48-1.00) for other diagnoses. Conclusions: Overall, STT in conjunction with a preliminary clinical history enables good agreement between trained assessors. Future work is required to evaluate its criterion validity. J Orthop Sports Phys Ther 2005;35:147-153.

Key Words: orthopedics, physical therapy, tests

1

Research Fellow, Teesside Centre for Rehabilitation Sciences, School of Health and Social Care, University of Teesside, Middlesbrough, UK. Professor, Teesside Centre for Rehabilitation Sciences, School of Health and Social Care, University of Teesside, Middlesbrough, UK. 3 Advanced Physiotherapy Practitioner, Middlesbrough Primary Care Trust, The Cleveland Health Centre, Middlesbrough, UK. This study was supported by the Research and Development Subcommittee of the South Tees Community and Mental Health Trust, and approved by the Nursing and Paramedical Research Advisory Group. Conflict of interests: Nigel Hanchard and Meg Gilbert are fellows of the Cyriax Organization and the Society of Orthopaedic Medicine, respectively, and teach for those bodies. Address correspondence to Nigel Hanchard, Teesside Centre for Rehabilitation Sciences, School of Health and Social Care, University of Teesside, Borough Road, Middlesbrough TS1 3BA, UK. E-mail: [email protected] 2

Journal of Orthopaedic & Sports Physical Therapy

S

houlder pain accounts for 1.2% of consultations by general medical practitioners and, in due course, 30% of the patients concerned are referred for physical therapy.9,14 It is imperative that treatment interventions are effective if the impact of this common condition on patients, therapists, and health care providers is to be minimized. However, the causes of shoulder pain are not uniformly defined (diagnostically labeled), making the relevant research evidence difficult to interpret and hampering optimization of management.9 Selective tissue tension (STT), a systematic approach to the physical assessment and diagnosis of musculoskeletal soft-tissue disorders, originally described by Cyriax,3 offers potential for uniformity in the diagnostic labeling of shoulder pain. Widely taught by the Society of Orthopaedic Medicine (SOM) and other organizations, it is in common clinical use, simple and logical,5 and has face validity. In addition, Hollingworth et al12 have investigated its criterion validity to some extent and have shown that steroid injections are significantly more effective for shoulder pain when placed at anatomical sites (determined by STT) than at trigger points. How147

RESEARCH REPORT

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

Nigel C. A. Hanchard, MSc 1 Tracey E. Howe, PhD 2 Meg M. Gilbert, BSc(Hons) 3

ever, evidence concerning the reliability of STT is contradictory.6,10-12,14,17 Reported values of ␬, used as a measure of between-tester agreement, vary from –0.30 for diagnoses of acute bursitis by Liesdek et al14 to 0.875 for agreement across combined diagnostic categories by Pellechia et al.17 The former value denotes worse agreement than would be expected by chance, and the latter denotes ‘‘near perfect’’ agreement. This issue clearly requires clarification. To explore it further and, if possible, to reconcile the contradictions in the literature, we evaluated diagnostic agreement between an expert in STT and 3 other assessors who had been trained in the approach.

METHODS

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

Subjects All general medical practitioners in the Middlesbrough area, UK, were informed of the study and invited to refer suitable subjects or agree to their direct recruitment from the physiotherapy waiting list. Subjects were required to be 18 years old or over and have shoulder pain, but were excluded if a history of clotting disorders, steroid therapy, inflammatory arthritis, or radiotherapy, fracture, or surgery around the shoulder was evinced. Informed consent was obtained following written and verbal explanation. Particular emphasis was placed on the potential discomfort associated with repeated physical testing and the option of withdrawal at any stage. Fifty-seven subjects were initially enrolled. The study was approved by the South Tees Nursing and Paramedical Research Advisory Group, Middlesbrough, UK.

TABLE 1. Standardized selective tissue tension assessment. Active movement

Cervical Shoulder/shoulder girdle

Passive movement

Shoulder/shoulder girdle Glenohumeral

Maximal isoGlenohumeral metric action Elbow

Extension, rotations, side flexions, flexion Elevation through abduction, elevation through flexion Elevation through abduction; horizontal adduction (‘‘scarf test’’)* Lateral rotation, abduction, medial rotation Adduction, abduction, lateral rotation, medial rotation Flexion, supination,* extension

* Optional ‘‘accessory’’ test.

Each subject’s STT assessments took place on 1 day in a private examination room. Initially, a physiotherapist who was otherwise independent of the study took a complete clinical history. This was recorded on a standardized form. Specifically, the form documented the subject’s date of birth, age, gender,

occupation, sports, and hobbies. Also documented were the mode of onset, duration of symptoms, pain distribution (recorded on a body chart), pain intensity (recorded on a visual analogue scale [VAS]7), and pain behavior (including aggravating and easing factors). To identify any nonshoulder or nonmechanical pathologies, or coexisting conditions that might cause frailty, information was also sought and recorded about drug history, general health, problems with other joints, paraesthesia, and past medical history (especially major illnesses and surgery). The completed form was given to each of the assessors immediately before they undertook their physical examination. These forms, therefore, informed the physical examinations and diagnoses, as would be the case in clinical practice. The VAS and body chart also enabled a baseline to be set before commencing this potentially irritative procedure. The independent physiotherapist recorded an additional VAS and pain chart before each reassessment for comparison with this baseline and confirmed the subject’s willingness to continue. The subject then undressed down to the waist (brassieres were not removed) before undergoing a standardized STT assessment13 (Table 1) by each of 4 assessors, all blind to the others’ diagnoses and to that of the referring general medical practitioner. Assessors had the option of using universal goniometers. To minimize cumulative irritation or fatigue, the objective assessments were separated by 15-minute rests. Additionally, the sequence in which the assessors saw the subjects was randomized so as to reduce the possibility of such order effects influencing the outcome. Each subject’s full series of objective assessments was completed within 2 hours or the session was aborted and he or she was removed from the study.

148

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

Assessors The assessors were a general medical practitioner of 9 years’ standing, with 5 years’ experience as an SOM tutor and 2 as an SOM fellowship tutor, and 3 chartered physiotherapists with, respectively, 16, 20, and 26 years’ postregistration experience. The physiotherapists had completed part A (relating to the cervical spine and shoulder) of the 3-part SOM course in the diagnosis and treatment of the musculoskeletal soft tissues at the outset of the study, but had not undergone formal examination or received certification.

Procedure

Analysis On the basis of STT (Table 2) and history, assessors assigned each subject to 1 of 5 predetermined categories, namely rotator cuff lesion, bursitis, capsulitis, other diagnosis (eg, cervicogenic or acromioclavicular joint problems), or no diagnosis. Combinations of the first 4 were allowed. Using MedCalc Version 7.4.0 statistical software (MedCalc Software, Mariakerke, Belgium), data were analyzed by ␬ to quantify the respective agreement between the expert and each of the other 3 assessors, over and above what would be expected by chance, across

all of the diagnostic categories combined. Evaluating agreement in this way (ie, between the expert and each of the nonexperts in turn) was considered more informative than evaluating agreement overall using multiple ␬.15 Combined diagnoses were treated as dropouts, irrespective of the extent to which they did or did not agree with the paired assessor’s diagnosis. This approach was considered evenhanded and had the advantage of simplifying data presentation. Kappa values for agreement between the expert and each of the other 3 assessors were also calculated for the specific diagnostic categories. Combined diagnoses were again treated as dropouts and data were

TABLE 2. Interpretation of selective tissue tension at the shoulder. Interpretation: Detail Type of Test

Evaluate ‘‘contractile’’ tissues (including tendons)

Evaluate bursa

Pain, minor lesion (eg, Supraspinatus tendonitis); pain and weakness, partial tear; painless weakness, complete tear Infraspinatus Subscapularis Biceps -

Subacromial bursa

Pain, stretch or pinch of sensitive tissue

Shoulder joint capsule

Subacromial bursa

Acromioclavicular joint

Active movements

Active neck movements

Evaluate willingness to move Replicate subacromial impingement

-

-

Painful arc, impingement

Screen the cervical spine

Reproduction of the patient’s pain implicates the cervical spine

Subacromial structures: rotator cuff and subacromial bursa -

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

Characteristic presentation Abduction is positive; lateral rotation may be positive in addition Lateral rotation is positive Medial rotation is positive Elbow flexion and supination are both positive Signs may be present (possibly contradictory or too numerous to attribute to a single tendon, and often with pain on ’release’), absent, or inconsistent; elbow extension may hurt, by abutting humerus and coracoacromial arch Capsular pattern is present (greatest restriction of lateral rotation, less of abduction, and still less of medial rotation) Possible noncapsular pattern of pain and, possibly, restricted movement (eg, pain on medial rotation with slight restriction) Pain at the extremes of shoulder ranges: but horizontal adduction (‘‘scarf test’’) is typically the most positive sign Painful arc on elevation and/or lowering Reproduction of pain on 1 or more movements

149

REPORT

Passive movements Evaluate noncontractile tissues

Structure

RESEARCH

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

Maximal isometric actions

Interpretation of Positive Findings: Principle

Purpose

neous (Table 4), agreement being only fair to moderate for bursitis, but good for rotator cuff lesions and other diagnoses, and good to very good for capsulitis.

TABLE 3. Diagnostic agreement between expert and 3 other assessors (diagnostic categories pooled).

Expert and assessor 1 Expert and assessor 2 Expert and assessor 3

␬*

95% CI†

0.67 0.75 0.61

0.48 - 0.84 0.60 - 0.90 0.44 - 0.78

DISCUSSION

* Kappa. Interpretation of ␬ ⬍ 0.20, poor; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80, good; 0.81-1.00, very good. † CI, confidence interval.

dichotomized and entered into 2 × 2 tables for the computations. Kappa is zero where agreement is due to chance. Values less than 0.20 denote poor agreement, 0.21 to 0.40 fair agreement, 0.41 to 0.60 moderate agreement, 0.61 to 0.80 good agreement, and 0.81 to 1.00 very good agreement.1

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

RESULTS Of the subjects who volunteered to participate, 9 were excluded prior to induction, having received either local steroid injections or long-term anticoagulant therapy. Two were subsequently removed from the study because their STT assessment sequences exceeded the time limits. Two withdrew because of increased pain after their third physical examinations. Fifty-three subjects completed the study: 32 male (mean ± SD age, 51 ± 13 years) and 21 female (mean ± SD age, 57 ± 12 years). Of these, 3 had both shoulders assessed. Data were therefore recorded for 56 shoulders in total (30 left, 26 right). The figure shows the diagnoses made by the referring general medical practitioners (n = 25). The STT expert diagnosed 19 rotator cuff lesions, 17 capsulitis, 13 bursitis, 6 other diagnoses, and 2 combined diagnoses. All assessors reached a diagnosis for every subject and fewer than 7% of the total diagnoses were combined (15 from 224 assessments). Analysis of the remaining 93% of diagnoses showed ‘‘good’’ agreement between the expert and the other assessors when the diagnostic categories were pooled (Table 3). However, when the diagnostic categories were analyzed separately, the results were heteroge-

The 53 subjects who completed this study had been attributed 11 different diagnoses by their referring medical practitioners (Figure). Forty-six percent of the subjects were allotted diagnoses that did not define any specific anatomical structure or pathology but described symptoms or the general mode of onset (shoulder injury, shoulder sprain, and shoulder pain). Primarily, we evaluated agreement between an expert and 3 other trained assessors using STT, in conjunction with a complete clinical history, to diagnose shoulder pain. Three ␬ values were calculated for overall agreement across the pooled diagnostic categories, and 12 for agreement on specific diagnostic categories. One of the former and 8 of the latter had 95% CIs that included the near perfect (0.875) ␬ value for overall agreement previously reported by Pellechia et al.17 The 95% confidence intervals for the remainder fell short of this. It is unclear whether they overlapped Pellechia et al’s17 interval estimates, which were not reported, and this hampers meaningful comparison of those investigators’ results with those of the present study. Such differences as do exist may relate to the number of assessors participating in the studies (ie, 4 in the present study, as opposed to 2). Four is probably the more relevant number because, in many busy outpatient settings, 4 different clinicians might well assess the same patient over time. Possible discrepancies between Pellechia et al’s17 results and our own might also relate to the levels of the assessors’ technical and interpretative skills, and the homogeneity of these attributes. Both are potential determinants of agreement. Pellechia et al’s17 assessors were characterized as having ‘‘advanced accomplishment’’ in the approach, while, by contrast, only 1 of our assessors (the SOM tutor) was considered an expert. Our physiotherapists’ technical and interpretative skills were implied rather

TABLE 4. Diagnostic agreement between expert and 3 other assessors by specific diagnostic categories (dichotomized data). Diagnostic Category

Expert and assessor 1 Expert and assessor 2 Expert and assessor 3

Bursitis ␬ (95% CI)*

Capsulitis ␬ (95% CI)

Rotator Cuff Lesion ␬ (95% CI)

Other Diagnosis ␬ (95% CI)

0.35 (–0.03-0.73) 0.58 (0.29-0.87) 0.41 (0.10-0.72)

0.82 (0.65-0.99) 0.81 (0.64-0.99) 0.63 (0.40-0.86)

0.71 (0.49-0.93) 0.79 (0.61-0.96) 0.71 (0.49-0.93)

0.69 (0.35-1.00) 0.78 (0.48-1.00) 0.74 (0.45-1.00)

Abbreviations: ␬, kappa; CI, confidence interval. * Interpretation of ␬: ⬍ 0.20, poor; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80, good; 0.81-1.00, very good.

150

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

Bicipital tendinitis 4%

Acromioclavicular joint pain 2%

Fibrositis 2%

Shoulder injury 2% Shoulder sprain 6%

Supraspinatus tendinitis 2% Rotator cuff injury 4% Painful arc 4%

Shoulder pain 38%

RESEARCH

Capsulitis 18% FIGURE. Diagnoses of referring medical practitioners.

than assured. Moreover, the homogeneity of these attributes across all 4 assessors was uncertain. This reflects real-life clinical practice and may enhance the generalizability of our results. Possible discrepancies between Pellechia et al’s17 results and our own might also relate to the relative complexity of the conditions with which the assessors were presented. In Pellechia et al’s17 study, no diagnoses of subacromial bursitis were made, although the assessors were given this option, hinting at a patient population with less complex variants of shoulder pain. In our study, subacromial bursitis was a diagnosis often made, but relatively infrequently agreed upon; Liesdeck et al14 reported similar results. This is perhaps not surprising. Cyriax considered subacromial bursitis to be the most nebulous cause of shoulder pain.3,4 It may mimic or accompany lesions of the rotator cuff tendons, to which it is closely apposed; and the bursa’s position may render it vulnerable to pinching during movements (active and passive) as well as isometric actions. Pellecchia et al17 and Liesdek et al14 ranked agreement on capsulitis and rotator cuff lesions in the same order as the present study, but Liesdeck et al14 reported the smallest absolute values for ␬. Liesdeck et al14 also

reported a smaller ␬ value than our own for subacromial bursitis. Indeed, a negative value (–0.30) was reported for acute bursitis, indicating worse agreement than might be expected by chance. The questionable preparation of their assessors (the majority of whom were stated to be already familiar with the Cyriax method, and who only underwent 2 hours of training) may account for this. Taken in overview, these studies do appear to confirm that, as intuition would suggest, agreement varies both with the complexity of the presenting disorder and the level and homogeneity of assessor expertise. This argues that, seminal though STT is, competence in its use cannot be taken for granted and adequate training is a necessity. Our results suggest that levels of training experienced by our assessors were sufficient to yield acceptable levels of reliability with the STT method. The present study, in common with others discussed thus far,14,17 was pragmatic, requiring assessors to synthesize subjective and objective data—as in a clinical setting—into diagnoses. In contrast, Hayes and Peterson10,11 evaluated 2 assessors’ agreement on STT’s component tests. In a subgroup whose pain was stable between assessments, the median K value for

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

151

REPORT

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

Frozen shoulder 18%

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

passive movements (pain resistance sequence) was 0.818. For individual movements, ␬ was very variable, ranging from 0.333 (for internal rotation) to 0.889 (for glenohumeral abduction and horizontal adduction).10 With respect to agreement on pain or weakness on isometric actions, Hayes and Peterson11 reported a median ␬ of 0.37. For specific actions, ␬ ranged from 0.00 (for shoulder internal rotation) to 0.45 (for shoulder abduction). Hayes and Peterson11 specifically considered the possibility that shortcomings in assessor training may have contributed to the poor agreement on the interpretation of isometric actions, and also questioned whether subjects were stabilized optimally during testing. Both points are probably pertinent. With respect to the second, the lack of proximal (trunk) stabilization during isometric medial and lateral rotation (at variance with the SOM method13 used by our assessors) would likely have engendered submaximal and inconsistent force generation, with obvious implications for the reproducibility of force and pain. The manner in which the shoulder was positioned during the isometric actions may also have contributed a deleterious effect on agreement. Shoulder medial and lateral rotation, and elbow flexion and extension were tested with the shoulder in an estimated 20° to 30° of abduction, a position harder to replicate and probably likelier to recruit unwanted muscle activity than the adducted (arm-at-side) position recommended by the SOM13 and employed in the present study. Hayes and Peterson11 also raise the issue of testers’ strength; but while a certain strength threshold is required to adequately resist the subject’s attempt at movement, it is likely that the requisite level is inflated by inefficient technique. Our study achieved acceptable results even though strength variations were likely to have been present among our testers, especially considering that these were 4 in number, and 1 (the expert, whose results were compared with each of the others’) was male. It is possible that access to patients’ histories influenced our assessors’ interpretations of isometric tests, but it seems unlikely that this would have enhanced agreement. Indeed, Bertilson et al,2 who evaluated the impact of history on the interpretation of tests for neck and shoulder pain, found that agreement for isometric shoulder elevation and abduction actually reduced when the assessors had access to the history. Other recent studies throw little additional light on the reliability of STT. Relatively poor agreement was reported by de Winter et al,6 but they were inexplicit as to their assessors’ level of training and further hampered appropriate interpretation of complex presentations by oversimplifying Cyriax’s3 diagnostic categories. Winters et al20 classified shoulder complaints in general practice by using cluster analysis, reporting only 3 clusters of symptoms and signs on passive or active movement that were stable over 2 test occa-

sions, and none that, according to the authors, fitted specific patterns described by Cyriax. These results stand somewhat at variance not only from those of the present study, where a diagnosis was made in every case, but also from those reported by other workers,14,17 and may reflect a number of methodological factors. In particular, Winters et al’s20 interpretation of the reported patterns is questionable. One of these, in fact, does appear to satisfy Cyriax’s3 criteria for a capsular pattern (Table 1). Cyriax diagnostic labels cannot be confidently ascribed to the remaining 2 patterns, which were characterized by pain with respectively ‘‘little’’ and ‘‘slight to average’’ restriction of movement. But this should not be overstated, in view of the omission of isometric muscle tests—an integral part of diagnosis by STT2 —from the assessment protocol.20 Considerations of STT’s reliability aside, its criterion validity remains to be conclusively established. This would require comparison with a criterion standard; but determination of an appropriate criterion is problematic. In relation to rotator cuff lesions, for example, even arthroscopy does not enable visualization of intratendinous lesions, which have been reported to account for between 8% and 26% of all partial thickness tears.8 Ultrasonography is valuable for detecting full thickness tears, but less so for small tears.8,19,21 Furthermore, asymptomatic tears are common,16,18 so that lesions correctly diagnosed by imaging modalities are of uncertain significance. The theoretical advantage of STT in this respect is that it reproduces symptoms or functional impairments, and therefore has implicit relevance. Conversely, local anaesthesia aims to abolish symptoms and (to an extent) functional impairments, should perhaps be considered as a criterion standard for evaluating STT’s validity in future research.

152

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

CONCLUSION In conjunction with a complete history, STT, as originally described by Cyriax,3 and with only minor refinements of technique,13 enables good diagnostic agreement between assessors who have been adequately trained in the approach. Variations in skill, technique, and the interpretation of STT may account for the apparently contradictory literature. These results have important implications for clinical practice and research, providing a basis for uniformity in both settings. We recommend that further work be undertaken into STT’s validity, using local anaesthesia as a criterion standard.

ACKNOWLEDGEMENT We thank the subjects who participated in this study, the assessors, and the service managers who made resources available to us.

153

Journal of Orthopaedic & Sports Physical Therapy® Downloaded from www.jospt.org at on January 18, 2017. For personal use only. No other uses without permission. Copyright © 2005 Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.

REPORT

J Orthop Sports Phys Ther • Volume 35 • Number 3 • March 2005

REFERENCES

RESEARCH

1. Altman DG. Practical Statistics for Medical Research. London, UK: Chapman & Hall; 1991. 2. Bertilson BC, Grunnesjo M, Strender LE. Reliability of clinical tests in the assessment of patients with neck/ shoulder problems-impact of history. Spine. 2003;28:2222-2231. 3. Cyriax JH. Illustrated Manual of Orthopaedic Medicine. 2nd ed. Oxford, UK: Butterworth Heinemann; 1993. 4. Cyriax JH, Cyriax PJ. Textbook of Orthopaedic Medicine Volume 1: Diagnosis of Soft Tissue Lesions. 8th ed. London, UK: Ballié`re Tindall; 1982. 5. Davies I. Cyriax techniques in orthopaedic medicine. Rheumatol Pract. 1997;4:21. 6. de Winter AF, Jans MP, Scholten RJPM, Deville W, van Schaardenburg D, Bouter LM. Diagnostic classification of shoulder disorders: interobserver agreement and determinants of disagreement. Ann Rheum Dis. 1999;58:272-277. 7. Finch E, Brooks D, Stratford PW, Mayo NE. Physical Rehabilitation Outcome Measures: A Guide to Enhanced Clinical Decision Making. 2nd ed. Baltimore, MD: Lippincott, Williams & Wilkins; 2002. 8. Fukuda H. The management of partial-thickness tears of the rotator cuff. J Bone Joint Surg Br. 2003;85:3-11. 9. Green S, Buchbinder R, Glazier R, Forbes A. Interventions for shoulder pain. Cochrane Database Syst Rev. 2000;CD001156. 10. Hayes KW, Petersen CM. Reliability of assessing endfeel and pain and resistance sequence in subjects with painful shoulders and knees. J Orthop Sports Phys Ther. 2001;31:432-445. 11. Hayes KW, Petersen CM. Reliability of classifications derived from Cyriax’s resisted testing in subjects with painful shoulders and knees. J Orthop Sports Phys Ther. 2003;33:235-246.

12. Hollingworth GR, Ellis RM, Hattersley TS. Comparison of injection techniques for shoulder pain: results of a double blind, randomised study. Br Med J (Clin Res Ed). 1983;287:1339-1341. 13. Kesson M, Atkins E. Orthopaedic Medicine: A Practical Approach. Oxford, UK: Butterworth Heinemann; 1998. 14. Liesdek C, van der Windt DAWM, Koes B, Bouter LM. Soft tissue disorders of the shoulder: a study of interobserver agreement between general practitioners and physiotherapists and an overview of physiotherapeutic treatment. Physiotherapy. 1997;83:12-17. 15. Maclure M, Willett WC. Misinterpretation and misuse of the kappa statistic. Am J Epidemiol. 1987;126:161169. 16. Miniaci A, Dowdy PA, Willits KR, Vellet AD. Magnetic resonance imaging evaluation of the rotator cuff tendons in the asymptomatic shoulder. Am J Sports Med. 1995;23:142-145. 17. Pellecchia GL, Paolino J, Connell J. Intertester reliability of the Cyriax evaluation in assessing patients with shoulder pain. J Orthop Sports Phys Ther. 1996;23:3438. 18. Sher JS, Uribe JW, Posada A, Murphy BJ, Zlatkin MB. Abnormal findings on magnetic resonance images of asymptomatic shoulders. J Bone Joint Surg Am. 1995;77:10-15. 19. Sonnabend DH, Hughes JS, Giuffre BM, Farrell R. The clinical role of shoulder ultrasound. Aust N Z J Surg. 1997;67:630-633. 20. Winters JC, Groenier KH, Sobel JS, Arendzen HH, Meyboom-de Jongh BM. Classification of shoulder complaints in general practice by means of cluster analysis. Arch Phys Med Rehabil. 1997;78:1369-1374. 21. Zehetgruber H, Lang T, Wurnig C. Distinction between supraspinatus, infraspinatus and subscapularis tendon tears with ultrasound in 332 surgically confirmed cases. Ultrasound Med Biol. 2002;28:711-717.