Neck or shoulder? A 23-year-old rule passes an independent test

Hi,

Welcome to the first Rule It In: five minutes a month on how to examine, and how to decide.

The study: the Wainner cluster holds up

In 2003, Wainner and colleagues studied 82 patients with suspected cervical radiculopathy or carpal tunnel syndrome. They found four examination findings that, together, did most of the work:

1. A positive upper limb tension test A (median-nerve bias).
2. Cervical rotation of less than 60° towards the painful side.
3. Symptoms relieved by neck distraction.
4. A positive Spurling's test.

With three of the four positive, the likelihood ratio was 6.1 (post-test probability about 65%). With all four, it was 30.3 (about 90%), though the confidence intervals were wide.

Clinical prediction rules often shrink when someone else tests them. This one mostly didn't. Grondin and colleagues (2026) tested it in a single neurosurgery centre. They studied 85 patients (from 109 consecutive referrals with neck pain, arm symptoms or both), about a third of whom had radiculopathy. The diagnosis was made on MRI and clinical assessment by a neurosurgeon blinded to the test results.

• 3 of 4 positive: specificity 91%, LR+ 7.7 (95% CI 3.2–20.7). That's slightly stronger than the original.
• All 4 positive: specificity 100%, but fewer than 1 in 5 patients with radiculopathy had all four.

At the bedside: don't wait for all four. Three positive findings make radiculopathy likely. For ruling out, be careful: in Wainner's study the tension test was positive in 97% of those with radiculopathy, but in the 2026 study it was positive in only 59%, so one negative test does not clear the neck. What did rule it out there was all four cluster findings being negative.

Caution: this is one centre with a modest number of cases, and the confidence intervals are wide. Wainner's patients were checked with nerve tests rather than MRI, so the two studies are not quite like for like; and the 2003 figures are quoted as they are commonly reported, because we could not obtain the full paper.

Test of the month: the arm squeeze test

What it asks: is this "shoulder" pain actually from the neck?

How: squeeze the middle third of the upper arm (over biceps and triceps) with moderate pressure, and compare it with the same pressure over the AC joint and the subacromial area.

Positive: the arm squeeze scores at least 3 points higher on a 0–10 pain scale than either shoulder site.

The numbers (Gumina 2013): 305 patients with cervical root compression were compared with 1,262 with shoulder conditions. Sensitivity was 96% and specificity 91–100%. Taking the lowest specificity gives LR+ 10.7 and LR− 0.04, matching the paper's weakest comparison.

What that means in the room: say the arm pain could be neck or shoulder, and before you squeeze you'd call the neck about a one-in-three chance. A clearly positive arm squeeze makes the neck the likely answer (about five in six). A clearly negative one makes the neck unlikely (about one in fifty), with one caution: the study compared groups whose diagnoses were already confirmed, a design that flatters a test. Even if you started out doubtful, a positive squeeze should send you back to the neck.

The catch: one study, not yet widely replicated. The paper's own likelihood ratios range from 10.6 (against glenohumeral arthritis, its weakest comparison) to 48 (against calcific tendinitis), with about 0.04 for a negative; we quote the weakest. Use it alongside the neck examination, not instead of it.

A or B? A painful cuff, or a torn one?

Neer's and Hawkins–Kennedy tests fire in both, so they can't separate them. What separates a full-thickness tear is a painful arc, a positive drop-arm and weak external rotation together.

In Park and colleagues' study of 552 patients undergoing arthroscopy, three findings did most of the work:

• a painful arc
• a positive drop-arm test
• weak external rotation

With all three positive, the chance of a full-thickness tear was 91%; with all three negative, 9%. That's from a starting point of 39%.

A physio clinic sees fewer tears than a surgical list. Starting from 30%, three positives still take you to about 87% (our calculation).

Famous but feeble: Speed's test

It's taught everywhere, but for SLAP lesions, pooled data give LR+ 1.67 and LR− 0.91 (Gismervik 2017). The confidence interval on its diagnostic odds ratio includes "no effect". A positive Speed's test barely changes anything. Use it to reproduce biceps-groove pain if you like, but not to decide.

One thing to do on Monday

Anyone with "shoulder" pain that spreads below the elbow gets a 60-second neck screen before you commit to a shoulder diagnosis: upper limb tension test A, Spurling's and the arm squeeze.

---

Sources

1. Wainner RS, Fritz JM, Irrgang JJ, et al. Reliability and diagnostic accuracy of the clinical examination and patient self-report measures for cervical radiculopathy. Spine. 2003;28(1):52–62.
2. Grondin F, Cook C, Hall T, Maillard O, Perdrix Y, Freppel S. An independent validation of a clinical prediction rule for the diagnosis of cervical radiculopathy with radicular pain. Braz J Phys Ther. 2026;30(3):101581. doi:10.1016/j.bjpt.2026.101581
3. Gumina S, Carbone S, Albino P, et al. Arm Squeeze Test: a new clinical test to distinguish neck from shoulder pain. Eur Spine J. 2013;22(7):1558–1563.
4. Park HB, Yokota A, Gill HS, et al. Diagnostic accuracy of clinical tests for the different degrees of subacromial impingement syndrome. J Bone Joint Surg Am. 2005;87(7):1446–1455. doi:10.2106/JBJS.D.02335
5. Gismervik SØ, Drogset JO, Granviken F, et al. Physical examination tests of the shoulder: a systematic review and meta-analysis of diagnostic test performance. BMC Musculoskelet Disord. 2017;18(1):41.

---

Rule It In is published by Trokos (a registered business name of The Trustee for Tompson Trust, ABN 38 360 148 648, Queensland) and written by Shane Tompson, physiotherapist. It's drafted with AI assistance, and every figure is checked against its source wherever the paper can be obtained. It's for clinicians, and supports but doesn't replace clinical judgement.

Updated 27 September 2026 with corrections to the version first emailed.

Get the next issue

All issues · trokos.com.au