Is a clinical exam reliable in diagnosing shoulder impingement?

Journal of Family Practice, July, 2008

No. Because the clinical examination of joints is highly dependent on examiner skill, the accuracy of these maneuvers outside a specialty clinic is likely to be even less reliable.

Silva L, Andreu JL, Munoz P, et al. Accuracy of physical examination in subacromial impingement syndrome. Rheumatology (Oxford). 2008;47:679-683.

Level of evidence

2b: Individual cohort study or low-quality randomized controlled trials (<80% follow-up)

Thirty consecutive adult patients, 24 to 82 years of age (average, 55 years), referred to a rheumatology clinic for evaluation of new-onset shoulder pain underwent a series of clinical examination maneuvers and MRI within 3 days of the examination. Their symptoms had lasted an average of 98 days.

The investigators excluded patients with shoulder trauma or surgery, inflammatory rheumatic diseases, painful cervical motion, or any other upper limb problems. These exclusions limit the ability to determine how well the clinical maneuvers distinguish among various causes of shoulder pain. The radiologist interpreting the MRI was experienced and was unaware of the patient's clinical history or examination findings. All examinations in this study were conducted by 1 examiner. It would have been nice to have more than 1 examiner to help determine inter-examiner reliability.

The authors compared several clinical maneuvers, with MRI serving as the gold standard. MRI, however, is a very questionable gold standard. Only 1 patient was unable to complete the MRI. Of the remaining 29 patients, 19 (65.5%) showed MRI signs of subacromial impingement and 15 (51.7%) had MRI evidence of subacromial or subdeltoid bursitis.

All the clinical maneuvers detected more than half the impinged shoulders (range, 58 %-79%) but had highly variable specificity (10%-60%). The overall diagnostic accuracy ranged from 45% to 66%. In other words, the diagnostic ability of the clinical maneuvers was not accurate.

Finally, the positive likelihood ratios ( LRs) ranged from 0.7 to 1.4 and the negative LRs (-LRs) ranged from 0.5 to 2.6. When a LR is 1 it gives no useful information. Similarly, the clinical examination in bursitis was also not reliable, with LRs ranging from 0.5 to 3.3 and the -LRs from 0.1 to 2.3. The Gerber test was the only maneuver accurate enough to rule out bursitis (-LR=0.1) and was only modestly accurate for diagnosis ( LR=3.3).

STUDY DETAILS

Design Cross-sectional

Funding Unknown/not stated

Setting Outpatient (specialty)

FAST TRACK

Diagnostic accuracy of clinical maneuvers to evaluate shoulder impingement ranged from 45% to 66%