Physical examination tests of the shoulder: a systematic review and meta-analysis of diagnostic test performance

Abstract

Background: Physical examination tests of the shoulder (PETS) are clinical examination maneuvers designed to aid the assessment of shoulder complaints. Despite more than 180 PETS described in the literature, evidence of their validity and usefulness in diagnosing the shoulder is questioned.

Methods: This meta-analysis aims to use diagnostic odds ratio (DOR) to evaluate how much PETS shift overall probability and to rank the test performance of single PETS in order to aid the clinician's choice of which tests to use. This study adheres to the principles outlined in the Cochrane guidelines and the PRISMA statement. A fixed effect model was used to assess the overall diagnostic validity of PETS by pooling DOR for different PETS with similar biomechanical rationale when possible. Single PETS were assessed and ranked by DOR. Clinical performance was assessed by sensitivity, specificity, accuracy and likelihood ratio.

Results: Six thousand nine-hundred abstracts and 202 full-text articles were assessed for eligibility; 20 articles were eligible and data from 11 articles could be included in the meta-analysis. All PETS for SLAP (superior labral anterior posterior) lesions pooled gave a DOR of 1.38 [1.13, 1.69]. The Supraspinatus test for any full thickness rotator cuff tear obtained the highest DOR of 9.24 (sensitivity was 0.74, specificity 0.77). Compression-Rotation test obtained the highest DOR (6.36) among single PETS for SLAP lesions (sensitivity 0.43, specificity 0.89) and Hawkins test obtained the highest DOR (2.86) for impingement syndrome (sensitivity 0.58, specificity 0.67). No single PETS showed superior clinical test performance.

Conclusions: The clinical performance of single PETS is limited. However, when the different PETS for SLAP lesions were pooled, we found a statistical significant change in post-test probability indicating an overall statistical validity. We suggest that clinicians choose their PETS among those with the highest pooled DOR and to assess validity to their own specific clinical settings, review the inclusion criteria of the included primary studies. We further propose that future studies on the validity of PETS use randomized research designs rather than the accuracy design relying less on well-established gold standard reference tests and efficient treatment options.

Keywords: Clinical test; Diagnosis; Meta-analysis; Physical examination; Rotator cuff tear; SLAP (superior labral anterior posterior) lesion; Shoulder; Shoulder pain; Subacromial impingement; Systematic review.

Fig. 1

The flow of the search and selection process in this systematic review and meta-analysis of physical examination tests of the shoulder. ¹QUADAS was scored for the all the articles that met the initial eligibility criteria. QUADAS-quality assessment tool for diagnostic accuracy studies

Fig. 2

Risk of bias in the 104 articles assessed by QUADAS

Fig. 3

a Evidence for validity of PETS in diagnosing SLAP lesions. The diamond represents a pooled DOR of 1.38 with a 95% confidence interval of [1.13, 1.69]. The Forrest plot also visualizes that the variation in performance between the presumably different PETS was low. Heterogeneity chi-squared was 26.6 (d.f. = 19), p = 0.12; I-squared (variation in DOR attributable to heterogeneity) was 28.5%. PETS-physical examination tests of the shoulder, DOR-diagnostic odds ratio. b Funnel plot of 2 × 2 tables constructed for SLAP lesions. Nos. 15, 17 and 19 were omitted in the meta-analysis due to outlier characteristics; i.e. visual outlier appearance (No. 19), Cooks distance (No. 19) and disease prevalences (for the 10 PETS) deviating from the average 46% (72% for Nos. 15 and 17 and 31% for No. 19). Assessment of spectrum effects showed that Nos. 19 (Biceps load II test, (Kim, S.H -01)) and Nos. 15 and 17 (the O’Brien and Crank test, (Myers, T.H -05)) had included a non-representative spectrum of patients; they had low average ages (30.6 years [No. 19] and 23.9 years [Nos. 15&17]) and for Nos. 15&17 only athletes younger than 50 were included. Ln(DOR)-natural logarithmic transformation of diagnostic odds ratio