Condition guide · Neck and upper limb
Cervical Radiculopathy
A 2025 trial compared surgery against non-surgical care twice over, in two separate groups. Surgery won one and drew the other — and which one you are in depends on what is pressing on the nerve.
In plain words. A nerve root leaving the neck becomes irritated or compressed, and the symptoms appear where that nerve travels — down the arm, into the forearm, often into specific fingers. That is why the problem is felt in the arm while the cause sits in the neck.
Two things commonly press on the nerve, and they behave differently. A disc herniation is usually a sudden event in a younger neck and often settles. Spondylosis — age-related narrowing from bone and disc change — develops slowly and does not resolve. The 2025 evidence suggests that distinction should change the advice you are given.
Key points
- The cause matters more than the diagnosis. In a 2025 trial of disc herniation, surgery beat non-surgical treatment on neck disability at 12 months by 7.4 points (95% CI 1.6 to 13.3). In a parallel trial of spondylosis, there was no significant difference (2.3 points, 95% CI -4.9 to 9.6). [1]
- Crossover tells its own story. Two patients in the non-surgical disc herniation group went on to have surgery, against eleven in the spondylosis group. [1] The group where surgery did not prove superior is the group where more people ended up having it.
- No serious adverse events were reported in either trial. [1]
- The conservative evidence base is weak and honest about it. A review of 59 trials and 4,108 participants could not pool its results and rated almost everything very-low certainty. [2]
- Some individual components do show effects. A 2025 network meta-analysis found neurodynamic techniques (SMD -1.45), articular treatment (-0.72) and cervical traction (-0.66) each reduced pain, with a combination rated moderate confidence. [3]
- Spurling's test confirms; it does not screen. Pooled specificity 0.92, sensitivity 0.53 — a negative test does not rule the condition out. [4]
What is actually happening
Eight pairs of nerve roots leave the spinal cord through the neck, each carrying sensation from and power to a defined territory in the shoulder, arm and hand. When one is compressed or inflamed at the point where it exits, the symptoms appear along that territory rather than at the site of the problem.
Two mechanisms dominate. In a disc herniation, disc material displaces backwards and contacts the nerve root; this tends to happen in younger necks, often relatively suddenly, and the herniation itself frequently reduces over time. In cervical spondylosis, the space the nerve travels through narrows gradually as discs lose height and bone spurs form; this is age-related, develops over years, and does not reverse.
Both produce arm symptoms that can look identical in clinic. They differ in natural history, and the 2025 trial data suggest they differ in how much surgery has to offer. [1]
Worth separating from both: age-related change visible on a scan is extremely common in people with no symptoms at all. Finding spondylosis on imaging does not, by itself, establish that it is causing your arm pain.
What it feels like
- Pain travelling from the neck into the shoulder blade, arm or hand, often in a band or stripe rather than the whole limb.
- Pins and needles or numbness in particular fingers — which fingers helps identify the level involved.
- Weakness in specific movements: gripping, lifting the arm, straightening the elbow.
- Symptoms worsened by extending or rotating the neck towards the painful side, and often eased by resting the hand on top of the head.
- Night pain and difficulty finding a comfortable sleeping position.
Arm pain that is worse than the neck pain is characteristic. If the neck hurts far more than the arm and nothing travels below the elbow, the problem may be mechanical neck pain instead.
How it is diagnosed
Diagnosis rests on the pattern — where the symptoms travel, which fingers are affected, which movements provoke them — supported by examination of sensation, power and reflexes in the arm. Clinical tests contribute, but their performance is worth knowing precisely, because it determines what a result actually means.
The nerve is irritated where it leaves the neck; the symptoms appear along the territory it supplies. Which fingers are involved is one of the more useful pieces of information in the whole assessment.
| Test | Sensitivity | Specificity | What that means in practice |
|---|---|---|---|
| Spurling's test | 0.53 | 0.92 | A positive result is meaningful; a negative one does not rule the condition out. Adding neck rotation or extension raised sensitivity to 0.67. [4] |
| Upper limb neurodynamic test (ULNT1) | 0.69 | 0.54 | Only fair accuracy overall, on low-certainty evidence. Recommended as an add-on to the existing pathway rather than a stand-alone test. [5] |
The practical reading: no single test settles this. Spurling's is useful for confirmation and poor for screening, [4] and the neurodynamic test adds information without being decisive on its own. [5] A clinician relying on one positive test is over-reading it, and one who excludes the diagnosis on a negative Spurling's is making the more common mistake.
Imaging is not needed to make the diagnosis in most cases and is generally reserved for when symptoms persist, when surgery is being considered, or when something in the history suggests another cause.
How physiotherapy and rehabilitation help
Most cervical radiculopathy improves without surgery, and non-surgical care is normally where treatment starts. What physiotherapy offers is symptom relief while that natural improvement happens, restoration of movement and strength in the arm and neck, and — where recovery does not come — a clear point at which to reconsider.
It is worth being straightforward about the state of this evidence. A systematic review of 59 trials with 4,108 participants found the studies too clinically and statistically heterogeneous to pool, synthesised them narratively, and rated the evidence very-low certainty for acupuncture, prednisolone, cervical manipulation and low-level laser therapy, and low to very-low for multimodal interventions. Its own conclusion is that the lack of high-quality evidence limits any meaningful conclusions. [2] That is the honest baseline.
Against that, a 2025 network meta-analysis of 36 trials comparing 25 interventions found measurable effects for individual components: neurodynamic techniques (SMD -1.45, 95% CI -1.88 to -1.02), articular treatment (-0.72, 95% CI -1.29 to -0.15) and cervical traction (-0.66, 95% CI -1.08 to -0.25). Its most effective combination — articular treatment, analgesic electrotherapy, neurodynamic techniques, strengthening exercise and cervical traction together — carried a moderate confidence rating. [3] Dry needling produced the largest single estimate (-3.40) but appeared in too few comparisons for the authors to draw firm conclusions, and they say so.
What a programme involves
| Component | What it is for | Evidence |
|---|---|---|
| Neurodynamic techniques | Restoring the nerve's ability to glide and tolerate tension as the arm and neck move. | Largest single effect SMD -1.45 (95% CI -1.88 to -1.02) in the network meta-analysis [3] |
| Manual therapy to the cervical and thoracic segments "articular treatment" |
Reducing pain and restoring segmental movement so the exercise becomes possible. | Supported SMD -0.72 (95% CI -1.29 to -0.15) [3] |
| Cervical traction | Reducing compressive load on the irritated nerve root. | Supported, modest SMD -0.66 (95% CI -1.08 to -0.25) [3]. One of the few places on this site where traction has supporting evidence |
| Strengthening exercise | Rebuilding neck and arm strength; the component most likely to matter once the acute pain settles. | Part of the best combination Included in the most effective combination, which carried a moderate confidence rating [3] |
| Analgesic electrotherapy | Short-term pain relief within a combined programme. | Part of the best combination One of the five components of the most effective combination [3] — stated here because the evidence says so, not because this site favours electrotherapy |
| Advice on positioning and load | Making the first fortnight tolerable and avoiding the postures that reproduce the arm pain. | Not separately tested Standard practice; not isolated as its own comparison in the cited reviews |
| Dry needling | Treating the muscular pain that accompanies the nerve symptoms. | Too uncertain to call Produced the largest single estimate (-3.40) but appeared in too few comparisons for the authors to draw firm conclusions, and they say so [3] |
| Acupuncture, cervical manipulation, low-level laser, prednisolone | Commonly offered for neck and arm pain. | Very low certainty Rated very-low certainty in the review of 59 trials, which could not pool its studies [2] |
An amber badge here mostly means the evidence is too weak or too thin to support a claim — not that the treatment fails. On this condition that caveat applies broadly: the largest review concluded that the lack of high-quality evidence limits any meaningful conclusions. [2]
What a course of treatment looks like
Assessment first, to confirm that the arm symptoms follow a nerve root rather than a shoulder or a peripheral nerve, to record what is affected — power, sensation, reflexes — and to establish a baseline that a later review can be compared against. Any progressive weakness is a reason to escalate rather than to continue.
Then a combined programme rather than a single technique, which is what the network analysis actually supports: manual therapy and neurodynamic work early for symptom relief, traction where it helps, and strengthening taking over as the acute pain settles. [3] Most cervical radiculopathy improves without surgery, so much of the early work is making the natural course tolerable while it happens.
The part that needs stating is the decision point. Non-surgical care is where treatment starts, but it should have a review date rather than running indefinitely. If arm pain, weakness or numbness is not improving, the source of the compression matters: where it is a disc herniation, surgery produced a clear advantage at twelve months, and where it is spondylosis it did not — though five times as many people in that second group crossed over to surgery anyway. [1] Those are different conversations, and knowing which one you are having is the point of the assessment.
When surgery is the question
This is where the 2025 evidence is genuinely useful, because it did not treat cervical radiculopathy as one condition. Two randomised trials ran in parallel: one in people whose nerve compression came from a disc herniation, one in people whose came from spondylosis. [1]
In the disc herniation trial, 87 patients had 12-month data. Surgery produced a mean Neck Disability Index difference of 7.4 points in its favour (95% CI 1.6 to 13.3, p=0.01). In the spondylosis trial, 88 patients had 12-month data and the difference was 2.3 points (95% CI -4.9 to 9.6, p=0.52) — not statistically significant. [1]
The crossover figures are worth as much as the primary result. In the disc herniation trial, two patients allocated to non-surgical treatment went on to have surgery. In the spondylosis trial, eleven did. [1] So in the group where surgery demonstrated no clear advantage, more than five times as many people crossed over to receive it. That is a reasonable description of how difficult spondylotic radiculopathy is to live with, and a reason to be cautious about reading "no significant difference" as "no need for surgery".
A separate 2024 systematic review comparing personalised multimodal physiotherapy against surgery found low-certainty evidence of no significant difference in arm pain or disability, but that physiotherapy improved significantly less on neck pain intensity, sensory loss and perceived recovery. It found no differences in numbness, range of motion, medication use or quality of life, reported no adverse events, and concluded that high-quality cost-effectiveness studies are still needed — cost-effectiveness was not actually assessed despite being the review's question. [6] It included only 117 participants across two original trials, which is the main reason to hold its conclusions loosely.
What the evidence supports — and what it does not
Supported
- Distinguishing disc herniation from spondylosis before advising on surgery — the two trials reached different answers. [1]
- Neurodynamic techniques, articular treatment and cervical traction as components with measurable effects on pain. [3]
- A combined programme of articular treatment, analgesic electrotherapy, neurodynamic techniques, strengthening and traction — moderate confidence. [3]
- Spurling's test for confirmation — specificity 0.92. [4]
Not supported
- Confident claims about any single conservative treatment. Very-low certainty across acupuncture, prednisolone, cervical manipulation and low-level laser therapy. [2]
- Using Spurling's test to rule the condition out. Sensitivity 0.53 — roughly half of those affected test negative. [4]
- Relying on the ULNT1 as a stand-alone test. Fair accuracy only, low certainty, recommended as an add-on. [5]
- Firm conclusions about dry needling. Largest single effect estimate in the network, but too few comparisons to conclude anything. [3]
- Claims about cost-effectiveness of physiotherapy versus surgery. The review asking that question did not assess it. [6]
How certain is this?
Evidence grade: Low to moderate, and the two halves of this page do not share a grade.
The surgical comparison is the stronger half. Two randomised trials, run in parallel with the same design and outcome measure, are a much better basis for the disc-versus-spondylosis distinction than indirect comparison would have been. [1] They are still modest in size — 87 and 88 patients with 12-month data — and a single trial programme from one country, so the effect sizes should be treated as provisional.
The conservative half is weaker, and the reviewers say so more bluntly than I would have. Fifty-nine trials could not be pooled, and the certainty ratings sit at very-low for most individual treatments. [2] The 2025 network meta-analysis is more encouraging, but network meta-analysis compares treatments largely through indirect chains of evidence, and its own confidence rating for the best combination is moderate rather than high. [3]
What survives is a shape rather than a set of numbers: most people improve without surgery; several physiotherapy components appear to reduce pain; the choice about surgery should depend on what is compressing the nerve; and nobody should be told any of this with more confidence than the evidence carries.
What to expect
Most cervical radiculopathy improves over weeks to months, and disc-related cases more reliably than spondylotic ones. Arm pain usually settles before numbness does, and numbness can persist after everything else has resolved without meaning that treatment failed.
Weakness is the symptom to watch. Mild, stable weakness that is improving is expected. Weakness that is worsening, or that affects the hand's fine movements, warrants prompt reassessment rather than waiting for the next appointment.
If symptoms are not improving after a reasonable course of non-surgical treatment, that is a point to reconsider rather than to persist indefinitely — particularly given how many people in the spondylosis trial ultimately chose surgery. [1]
Common questions
Do I need surgery?
Probably not immediately, and the answer depends on the cause. In the 2025 trials surgery was better than non-surgical treatment for disc herniation (7.4 NDI points at 12 months) but not significantly better for spondylosis. [1] Most people start with non-surgical care. The reasons to consider surgery sooner are progressive weakness, or symptoms that have not improved after a proper course of treatment.
My scan shows disc degeneration. Is that the cause?
Not necessarily. Age-related change in the cervical spine is very common in people with no symptoms at all, so finding it on a scan does not establish it as the source of your arm pain. The diagnosis rests on the pattern of symptoms and the examination, with imaging used to answer specific questions rather than to make the diagnosis.
My Spurling's test was negative. Does that rule it out?
No. Pooled sensitivity is 0.53, meaning roughly half of people with cervical radiculopathy will test negative. The test is highly specific (0.92), so a positive result is informative — but it is better suited to confirmation than to screening. [4]
Is traction worth doing?
Here, unusually, yes — with modest expectations. Cervical traction was one of the components associated with reduced pain in the 2025 network analysis (SMD -0.66, 95% CI -1.08 to -0.25), and part of the best-performing combination. [3] That is a narrower claim than traction is often given credit for, and it does not transfer to low back pain.
How is this different from carpal tunnel syndrome?
Both cause numbness and tingling in the hand, and they are genuinely confused. Carpal tunnel symptoms start at the wrist and typically affect the thumb, index and middle fingers, often waking people at night. Radiculopathy symptoms come from the neck, usually involve pain travelling down the arm, and change with neck position. See carpal tunnel syndrome. The two can also coexist.
Should I stop moving my neck?
No. Positions that ease the symptoms are worth using in the early painful phase, but prolonged immobility is not the treatment. The components with the best evidence — neurodynamic techniques, articular treatment, strengthening — all involve movement. [3]
References
- Taso M, Sommernes JH, Sundseth J, et al. Surgical versus Nonsurgical Treatment for Cervical Radiculopathy. NEJM Evidence. 2025 Apr;4(4):EVIDoa2400404. doi:10.1056/EVIDoa2400404 PMID 40130970 Randomised controlled trials
- Plener J, Csiernik B, To D, et al. Conservative Management of Cervical Radiculopathy: A Systematic Review. Clinical Journal of Pain. 2023 Mar 1;39(3):138–146. doi:10.1097/AJP.0000000000001092 PMID 36599029 Systematic review
- Núñez de Arenas-Arroyo S, Mavridis D, Martínez-Vizcaíno V, et al. What components and formats of rehabilitation interventions are more effective to reduce pain in patients with cervical radiculopathy? A Systematic review and component network meta-analysis. Clinical Rehabilitation. 2025 Oct;39(10):1296–1310. doi:10.1177/02692155251365193 PMID 40776625 Systematic review and network meta-analysis
- Lin LH, Lin TY, Chang KV, et al. Diagnostic Performance of Spurling's Test for the Assessment Subacute and Chronic Cervical Radiculopathy: A Systematic Review and Meta-analysis. American Journal of Physical Medicine & Rehabilitation. 2025 Aug 1;104(8):717–723. doi:10.1097/PHM.0000000000002707 PMID 39938056 Diagnostic accuracy meta-analysis
- Shen P, Chi-Chung Tsang R, Liang Y, et al. Diagnostic accuracy of the upper limb neurodynamic test with median bias (ULNT1) for cervical radiculopathy: a systematic review and meta-analysis. Physiotherapy. 2023 Sep;120:17–25. doi:10.1016/j.physio.2023.06.001 PMID 37356367 Diagnostic accuracy meta-analysis
- Klein Heerenbrink S, Coenen P, Coppieters MW, et al. (Cost-)effectiveness of personalised multimodal physiotherapy compared to surgery in patients with cervical radiculopathy: A systematic review. Journal of Evaluation in Clinical Practice. 2024 Oct;30(7):1227–1238. doi:10.1111/jep.14036 PMID 38825757 Systematic review
About this guide
If you need assessment
This page explains. It does not diagnose.
Arm pain with numbness or weakness needs examining rather than guessing at, partly to establish which nerve is involved and partly to exclude the other causes that look similar. Worsening weakness in the arm or hand should be assessed promptly. Physiotherapist India publishes information and takes no bookings. Ask a qualified physiotherapist to examine your own case before acting on anything written here.