Intended for healthcare professionals

Clinical Review State of the Art Review

Advances in the diagnosis and management of neck pain

BMJ 2017; 358 doi: (Published 14 August 2017) Cite this as: BMJ 2017;358:j3221
  1. Steven P Cohen, professor1 2,
  2. W Michael Hooten, professor3
  1. 1Departments of Anesthesiology and Critical Care Medicine, Neurology and Physical Medicine and Rehabilitation, Johns Hopkins School of Medicine, Baltimore, MD 21029, USA
  2. 2Uniformed Services University of the Health Sciences, Bethesda, MD 20889 USA
  3. 3Department of Anesthesiology and Perioperative Medicine, Mayo Clinic School of Medicine, Rochester, MN 55905 USA
  1. Correspondence to: S P Cohen scohen40{at}


Neck pain imposes a considerable personal and socioeconomic burden—it is one of the top five chronic pain conditions in terms of prevalence and years lost to disability—yet it receives a fraction of the research funding given to low back pain. Although most acute episodes resolve spontaneously, more than a third of affected people still have low grade symptoms or recurrences more than one year later, with genetics and psychosocial factors being risk factors for persistence. Nearly half of people with chronic neck pain have mixed neuropathic-nociceptive symptoms or predominantly neuropathic symptoms. Few clinical trials are dedicated solely to neck pain. Muscle relaxants and non-steroidal anti-inflammatory drugs are effective for acute neck pain, and clinical practice is mostly guided by the results of studies performed for other chronic pain conditions. Among complementary and alternative treatments, the strongest evidence is for exercise, with weaker evidence supporting massage, acupuncture, yoga, and spinal manipulation in different contexts. For cervical radiculopathy and facet arthropathy, weak evidence supports epidural steroid injections and radiofrequency denervation, respectively. Surgery is more effective than conservative treatment in the short term but not in the long term for most of these patients, and clinical observation is a reasonable strategy before surgery.


Neck pain is a common condition and a leading cause of disability worldwide.12 Despite the enormous burden that neck pain exacts on society, it attracts only a fraction of the research money and publicity given to back pain.3 In this article, we explore the epidemiology, diagnosis, and treatment of neck pain. Emphasis is given to controversial topics and treatments that are most commonly used and investigated (such as integrative and interventional treatments), with particular attention paid to areas that are most relevant to academics and specialists.

Sources and selection criteria

In February 2017, we searched the Medline database, Embase, Google Scholar, and the Cochrane Database of Systematic Reviews using the search terms “cervical pain”, “neck pain”, “cervical radiculopathy”, “cervical radicular pain”, and “cervical myelopathy”, with no restrictions. For individual sections, key words relating to the relevant topics (for example, facet joint, epidural steroid injections, physical examinations, antidepressants, acupuncture, and surgery) were identified and cross referenced with the initial search terms using the above databases. We considered animal and experimental studies, systematic and other reviews, meta-analyses, clinical trials, and for certain sections in which higher grade evidence was lacking (such as treatment and complications) case reports and case series. Systematic reviews and clinical trials, particularly larger randomized studies, were prioritized over lower grade evidence. We also obtained additional articles by examining reference lists.


Neck pain has a high prevalence in developed countries. One systematic review estimated mean point, annual, and lifetime prevalence rates of 7.6% (range 5.9-22.2%), 37.2% (range 16.7-75.1%), and 48.5% (range 14.2-71%), respectively.1 According to the Global Burden of Disease 2010 study, neck pain is the fourth most common cause of disability in the United States, after back pain, depression, and other musculoskeletal disorders.2 Women are more likely to experience neck pain, with peak prevalence occurring in middle age.14 A recent review estimated that the annual cost of low back and neck pain was $87.6bn (£67.8bn; €77.2bn) in the US, ranking third behind diabetes and heart disease.5

Risk factors

Several risk factors predispose to the development of neck pain, including psychopathology, genetics, sleep problems, smoking, obesity, sedentary lifestyle, previous neck pain, trauma, back pain, and poor general health.6789 Sports and work injuries have also been associated with neck pain, with the highest incidence noted for race car driving, wrestling, and ice hockey.4 Although office and computer workers, manual laborers, healthcare workers, and occupational drivers are more likely than others to experience neck and shoulder pain, low job satisfaction and poorly perceived work support are the major work related factors associated with neck pain.10


Neck pain can be classified in several ways—for example, acute versus chronic or associated versus not associated with occipital headaches. However, the most relevant classification separates neck pain into neuropathic, which requires discrete identifiable nerve(s) injury as the cause of symptoms, and non-neuropathic, because this distinction affects diagnostic assessments and treatment at all levels of care. In the lumbar region, studies have reported prevalence rates of 16-55% for neuropathic pain,1112 with one review reporting an aggregate rate of 36.6%.13 The only study performed in the cervical region found that 43% of 100 patients had non-neuropathic pain, 7% had predominantly neuropathic pain, and 50% had mixed pain.14 The low proportion of patients with solely neuropathic pain probably resulted from the methodology used because selection criteria required patients to have neck pain (those with arm pain without neck pain were excluded). The high proportion of patients with mixed neuropathic-nociceptive pain was attributed to the fact that radicular pain is usually caused by degenerative conditions that predispose a person to nociceptive pain (for example, herniation occurring in a degenerative disc or facet hypertrophy resulting in foraminal stenosis).3 People with neuropathic pain had higher levels of functional impairment and psychopathology, similar to what has been reported for low back pain.11 Not surprisingly, those with neuropathic or mixed pain states were more likely to receive procedural interventions such as surgery and epidural steroid injections (ESIs).14

Natural course

Acute neck pain

Most episodes of acute neck pain resolve within two months, although about half of patients continue to have low grade symptoms or recurrences and to seek healthcare for their symptoms for more than one year.15161718 Ironically, neither early treatment nor radiographic degeneration seems to have a meaningful impact on prognosis.1619 Variables that predict persistence include female sex, older age, presence of radiculopathy, higher baseline pain intensity, multiple pain sites, smoking, obesity, poor general health, and a variety of psychosocial factors.151617192021

Cervical radiculopathy

Although cervical radicular pain is associated with a worse prognosis than axial pain, most patients improve.21 A retrospective study of 561 patients with cervical radicular pain reported that although recurrences were common (31.7%), 90.5% of people had minimal or no pain at a mean follow-up of 5.9 years.18 A systematic review evaluating the natural course of cervical radicular pain from a herniated disc found that although most patients improved within six months, complete improvement, which was seen in 83% of people, took two to three years. The main factor associated with poor outcome was an ongoing worker’s compensation claim.22 These findings are consistent with radiological studies showing that about half of cervical disc herniations will decrease within the first six months and about 75% will decrease by more than 50% within two years.23 Data on the natural progression of cervical stenosis are scant, but unlike disc herniations, the anatomical derangements of spinal stenosis do not regress with time.

Cervical myelopathy

Cervical myelopathy results from disease (for example, myelitis) or injury (for example, trauma or syrinx) affecting the spinal cord, which causes upper motor neuron signs. In young people, trauma is the most common cause of myelopathy, whereas spondylosis is the most common cause later in life.24 The incidence and point prevalence rates of cervical myelopathy are conservatively estimated to be 41 per million and 605 per million, respectively.25 The natural course of cervical myelopathy is highly variable and is characterized by periods of quiescence and stepwise progression.

Whereas myelopathy is sometimes considered an indication for surgery, studies comparing surgical and non-surgical outcomes have been mixed. A randomized trial comparing surgical to non-surgical treatment for mild to moderate spondylotic myelopathy reported that more than 80% of both groups exhibited no progression or improvement at three year follow-up.26 Another prospective study also described a 20% deterioration rate at a mean follow-up of three years.27 A prospective study in 62 patients found that about three quarters of surgically and non-surgically treated patients were satisfied with treatment at a mean follow-up of 11 months, although those managed medically experienced more neurological deterioration.28

Other retrospective studies have reported worse outcomes for spondylotic myelopathy. One found that 10 of 27 patients treated conservatively over six months experienced deterioration that necessitated surgery,29 whereas another found that all 22 patients with cervical myelopathy reported progression, with 21 proceeding to surgery.30 The authors of a literature review combined with a systematic review31 as well as a consensus statement on conservative management32 concluded that 20-60% of patients will deteriorate at three to six year follow-up, with only the area of compression weakly predicting disease progression. About 23% of patients with spondylotic cord decompression without myelopathic symptoms will progress to myelopathy within four years.32 Other factors associated with poorer prognosis with conservative management include small spinal canal area, greater cord compression, younger age, and more severe symptoms.2733


Many biological and psychosocial factors can predispose to chronic neck pain, but it is less well known that chronic pain is often associated with anatomical, psychological, social, and professional factors. This is consistent with the biopsychosocial model, which considers pain to be a dynamic interaction between biological, psychological, and social factors unique to each individual (fig 1). Consequently, researchers have examined whether a variety of interventions can prevent the development of neck pain and its transition to chronic pain and disability.


Fig 1 The biopsychosocial model of pain posits that biological, psychological, and social factors influence who develops chronic pain (pink circles) and that chronic pain has biological, psychological, and social consequences (blue circles). Courtesy of Frank Corl (Mayo Clinic), Steven Cohen, and W Michael Hooten

The risks and costs of using drugs and other interventions to prevent neck pain in asymptomatic high risk populations outweigh the benefits so this approach is not indicated. One meta-analysis found that education did not prevent neck and back pain.34 Another review of multiple controlled trials found that exercise was effective in preventing neck and low back pain but there was no evidence to support or refute the effectiveness of ergonomic or risk factor modification.35 In people with acute whiplash injury, one large randomized controlled study performed in 405 patients who presented to the emergency department found that an educational video added to usual care resulted in a 7.9% (95% confidence interval −2.0 to 17.8) decrease in the proportion of people who still had pain at 24 weeks compared with usual care alone.36


A thorough history and physical examination is needed to distinguish neuropathic pain from mechanical neck pain because treatment decisions are based on this distinction. Medical and systemic disorders can cause neck pain (box 1; fig 2), and most are regional musculoskeletal or neuropathic in nature.

Box 1: Uncommon medical and systemic causes of neck pain

  • Metastatic tumor

  • Multiple myeloma

  • Spinal cord tumors

  • Chordoma

  • Rheumatoid arthritis

  • Seronegative spondyloarthropathies

  • Osteomyelitis

  • Epidural abscess

  • Discitis

  • Herpes zoster

  • Meningitis


Arteriovenous fistula or malformation

  • Paget’s disease

  • Osteoporotic fractures

  • Peripheral neuropathy

  • Amyotrophic lateral sclerosis

  • Transverse myelitis

  • Guillain-Barré syndrome

  • Brachial plexus lesion


Fig 2 Short T1 inversion recovery (STIR) magnetic resonance image showing an osteochondroma causing spinal cord compression (A) in a patient with multiple hereditary exostoses. The patient presented with myelopathy and mild spinal cord signal changes. Arrow B indicates decreased signal within the cerebrospinal fluid representing a flow related artefact. Courtesy of Brandon Childers, Department of Radiology, Johns Hopkins

Suspected neuropathic pain

Nerve root compression caused by an acute intervertebral disc herniation may initially produce neck pain, followed by arm pain. A history of physical exertion or trauma is often assumed to be a causative factor but is present in less than 15% of patients.18 Radicular pain typically occurs in a dermatomal pattern, although more than half of people have multiple nerve root involvement and there is considerable dermatomal overlap.37 Clinical symptoms of nerve root compression may also be secondary to foraminal stenosis, which is typically associated with an insidious onset.38 Exacerbating factors can include coughing, sneezing, or other activities that increase subarachnoid pressure. Another common source of neuropathic pain is cervical spinal stenosis. Presenting symptoms can include neck pain, stiffness, and upper extremity radicular pain. One potential consequence of cervical spine stenosis is the development of cervical myelopathy, which is characterized by symptoms of upper motor neuron impairment.39 In patients with presumed cervical myelopathy, amyotrophic lateral sclerosis should be suspected when fasciculations and bulbar signs are evident.

Signs of nerve root compression can be identified on physical examination but may be obscured by the presence of active myofascial trigger points, which were found in one study to occur in 51.2% of people with cervical radiculopathy.40 On inspection, the head and neck may be slightly tilted towards the affected side. Muscles of the neck, shoulder, arm, and hand should be observed for atrophy, which may indicate the presence of long standing nerve root dysfunction. However, numbness in the hands and fingers is often indicative of carpal or cubital tunnel syndrome and not spinal nerve root compression. Physical examination maneuvers may help establish a diagnosis of nerve root compression. For example, Spurling’s test and upper limb tension tests have sensitivities exceeding 50%, and the specificities of Spurling’s test, shoulder abduction and neck distraction tests range between 80% and 90% (table 1).41424344 When performing physical examination maneuvers, shoulder pain elicited during rotation and abduction of the arm suggests a primary shoulder problem rather than nerve root compression. Physical examination findings are often non-specific for cervical stenosis but include reduced cervical range of motion and paraspinal tenderness. In general, tests may be more accurate in acute than in chronic radiculopathy and for herniated discs compared with stenosis, but the use of multiple tests increases precision.44

Table 1

Clinical and diagnostic evaluation of neck pain*341424344

View this table:

Diagnostic tests

Diagnostic imaging is recommended for the evaluation of neuropathic neck pain. Although magnetic resonance imaging (MRI) is superior to computed tomography for evaluating soft tissue abnormalities, computed tomography or computed tomography-myelography may help distinguish osteophytes from soft tissue abnormalities and can be used when MRI is contraindicated. For the evaluation of nerve root compression, MRI findings should be interpreted in the context of the clinical presenation.45 In an MRI study of 78 patients with cervical radiculopathy for less than one month, the clinically affected root was identified in 73% of patients.46 However, the false positive rate was 45% and the false negative rate was about 26%, indicating the absence of compression at the affected level. Regardless of the performance characteristics of imaging modalities, brachial plexus pathology should be considered in patients with upper extremity pain, weakness, sensory loss, and a non-diagnostic imaging study.

Electromyography and nerve conduction studies are the key electrodiagnostic tests used to identify physiological nerve root abnormalities. The two main objectives of electrodiagnostics are:

  • To confirm the existence of nerve root dysfunction and exclude other peripheral nerve disorders, such as plexopathy

  • To identify which nerve root(s) is involved and determine the type of nerve root dysfunction, such as demyelination, axonal loss, and conduction block.

The sensitivity of electrodiagnostic testing for cervical radiculopathy ranges from 50% to 71%,4748 so a non-diagnostic test result does not exclude nerve root dysfunction.49 In clinical practice, the utility of electromyography could be affected by confounding musculoskeletal disorders such as myofascial pain and lateral epicondylitis, which are often present in patients referred for electrodiagnostic testing.50

Evaluation of mechanical non-radicular neck pain

The onset of non-radicular neck pain is usually insidious, but pain referral patterns may help distinguish between common sources of pain. For example, axial neck pain referred to the occipital, suboccipital, shoulder, or mid-back regions is suggestive of cervical facet joint pain.51 Specifically, the atlanto-occipital, and atlanto-axial joints can be associated with occipital and posterior auricular pain, pain emanating from the C2-3 or C3-4 facet joints can extend into the occipital and suboccipital regions, C4-5 or C5-6 facet joint disease can cause pain radiating into the shoulder, and the C6-7 and C7-T1 joints typically refer pain to the mid-back and scapular regions.51 Cervical facet pain does not usually extend distal to the shoulder and is not associated with neurological deficits, which can help distinguish it from radicular pain. Axial neck pain associated with headaches, unilateral or bilateral shoulder pain, non-radicular arm pain, ocular and vestibular dysfunction, and anterior chest wall pain is suggestive of cervical discogenic pain.52 Activities that may provoke or aggravate discogenic pain include coughing, lifting, or applying pressure on the cervical spinous processes, whereas lying supine may relieve pain.52 Regardless of the underlying source, studies have shown increased myoelectric activity in patients with chronic neck pain,53 and the prevalence of trigger points in the neck musculature of people with chronic neck pain can exceed 50%.54 This suggests that myofascial pain is often associated with primary pain generators.

The abrupt onset of axial neck pain is often the result of trauma; exposure of the cervical spine, including its muscles and ligaments, to excessive forces; or an underlying medical condition. When trauma is suspected, the clinical history should focus on identifying the potential mechanism of injury. On physical examination, guarding may result in a reduced range of motion and the neck musculature may be tender to palpation. The absence of midline cervical spine tenderness, focal neurological deficits, intoxication, and painful distracting injury (clinically apparent pain that might distract the patient from the pain of a cervical spine injury), together with a normal level of consciousness indicates a low probability of serious cervical spine injury after blunt trauma.55

Whiplash refers to an “acceleration-deceleration mechanism of energy transfer to the neck” often stemming from rear or side impact motor vehicle collisions.56 Although neck pain after whiplash injury can be attributed to the facet joints in about 50% of people,57 other structures are important contributors to pain including the discs,5658 muscles,5659 and ligaments.5660 Consequently, pain associated with whiplash injury is often referred to the trapezius muscle, shoulder, mid-back, and, to a lesser extent, the face.61 Up to 80% of people will experience neck pain within one day of sustaining a whiplash injury and about 50% will continue to report neck pain one year after the initial injury.62 Although acute pain after whiplash may be secondary to subclinical disease (soft tissue injury or trauma involving the spine),56 no reliable link has been established to account for persistent chronic pain. Diagnostic imaging should include flexion-extension radiographs of the cervical spine to look for fractures or vertebral body malalignment; however, radiographs may yield limited diagnostic information.63 Depending on the extent of injury, computed tomography and MRI may be warranted to look for fractures and ligamentous injury,64 although studies have generally found no association between MRI findings and persistent pain after whiplash.65

Red flags

A key component of the clinical evaluation of neck pain is to identify serious pathology and non-musculoskeletal diseases that may be the source of pain and related symptoms. The differential diagnosis of neck pain is broad but includes trauma and non-musculoskeletal disease processes that can be classified as neoplastic, inflammatory, infectious, vascular, endocrinological, and neurologic in origin (box 1). Red flags refer to signs and symptoms that raise suspicion of something more serious than conventional musculoskeletal disorders, such as spinal cord injury, infection, tumors, or cardiovascular disease. In contrast to low back pain,66 the diagnostic accuracy of red flag findings for identifying life threatening or critical disease involving the neck has not been reported. In addition to trauma and other factors that predispose a person to dangerous neurological sequelae, such as atlanto-axial subluxation in Down’s syndrome and inflammatory arthritis,367 important risk factors and clinical features suggestive of a non-musculoskeletal disease process can be generally categorized as: age related, physical signs and symptoms, neurological findings, and serum markers of inflammation (figs 3-5).52


Fig 3 Red flags for neck pain necessitating further investigation. Courtesy of Frank Corl (Mayo Clinic), Steven P Cohen, and W Michael Hooten


Fig 4 T1 weighted sagittal magnetic resonance image showing increased distance between the posterior aspect of the anterior arch of C1 and the anterior margin of the dens of C2 (arrow), indicating atlanto-axial subluxation in a patient with rheumatoid arthritis. There is also anterior positioning of the posterior arch of C1 resulting in mild to moderate spinal cord compression without edema. Courtesy of John Carrino (Weil Cornell Medical College)


Fig 5 T2 weighted axial magnetic resonance image through C7-T1 showing a syrinx (arrow). Courtesy of John Carrino (Weill Cornell Medical College)

Causes and treatments

There are as many potential pain generators in the neck as there are anatomical structures, and in the absence of a pathognomonic means for diagnosis, it can be difficult to identify the source of pain. Many treatments are predicated on precise diagnoses, often in the form of “diagnostic blocks,” but in the absence of any confirmatory reference standard the validity of diagnostic injections can never truly be known. Given the interdependent association between adjacent structures, most people probably have multiple concurrent pain generators (fig 6).


Fig 6 Sagittal view of cervical spine showing potential pain generators. Courtesy of Frank Corl (Mayo Clinic)

Mechanical non-neuropathic causes

Non-specific neck pain

Few clinical trials have evaluated drugs for neck pain, so treatment is often based on generalization from studies performed for back pain. Because no study evaluating pharmacotherapy for non-neuropathic spinal pain has ever “diagnosed” a specific cause on the basis of injections (such as facet blocks), the study population is often referred to as having “non-specific” neck pain. This denotes pain that cannot be attributed to a specific cause often because a proper investigation has not been conducted. It is unclear whether drugs that are effective for non-specific spinal pain would yield similar benefit in a more homogeneous population, such as those with discogenic pain. Given that these drugs have small effect sizes even in ideal circumstances, it is reasonable to try non-pharmacological treatment options first.

Reviews have concluded that systemic non-steroidal anti-inflammatory drugs (NSAIDs) are effective in back pain, but they carry a risk of adverse effects in people over 60 years and have not been evaluated in neck pain. Two moderately sized placebo controlled trials established short term (2-8 days) benefit for topical diclofenac in people with neck and upper back pain secondary to suspected muscle and joint disease (mean difference 63% in pain scores in the treatment group versus 24% in the control group),6869 which is consistent with reviews showing efficacy for topical NSAIDs in acute and chronic musculoskeletal pain.70

Muscle relaxants can alleviate pain and improve function in patients with spinal pain and are more effective for acute pain than for chronic pain. Two randomized controlled trials (RCTs) evaluating cyclobenzaprine in 1405 patients with acute neck (more than a third of patients) or back pain secondary to muscle spasm found that intermediate doses (15 mg/d) and high doses (30 mg/d) were more effective than placebo (P<0.03), but that low doses (7.5 mg/d) were ineffective (responder rate for intermediate dose 11-20% higher than placebo group).71 A double blind crossover study in 90 people with joint or back pain, compared the anti-inflammatory drug benorylate alone to benorylate and chlormezanone. It found no benefit for add-on muscle relaxant therapy, although subgroup analysis found significant improvement in patients with neck pain for pain reduction and sleep.72 Comparative effectiveness research is one of the top chronic pain research priorities of the National Institutes of Health (NIH), yet few pain studies have compared treatments. A three arm randomized trial compared spinal manipulation; home exercise and advice; and pharmacotherapy with NSAIDs, acetaminophen, or muscle relaxants (or a combination of these drugs) in 272 people with acute and subacute neck pain. Patients in the manipulation and exercise groups fared statistically significantly better than those who received pharmacological treatment up to 12 month follow-up (50% responder rates 82%, 77%, and 69%, respectively, at week 12).73

Collectively, these studies provide moderate evidence that topical NSAIDs such as diclofenac are effective for acute and chronic neck pain and weak evidence supporting muscle relaxants for subacute neck pain associated with muscle spasm. However, neither drug is better than non-pharmacological alternative treatments.

Myofascial pain

Myofascial pain is a common cause of neck pain that involves discrete or diffuse areas of sensitivity within one or more muscle. The causes of myofascial pain are poorly understood, but muscle pain can develop secondary to biomechanical imbalances, trauma, emotional stress, and even endocrine and hormonal abnormalities. Tender muscles release excess acetylcholine, which can result in dysfunctional motor endplates, sustained muscle contractions, local ischemia, sarcomere shortening, and the release of inflammatory mediators, in what can devolve into a vicious circle.74

A hallmark of myofascial pain is the presence of palpable trigger points, which are taut muscle bands that refer pain in a defined pattern spontaneously or after stimulation, although their clinical relevance is controversial. In the neck, primary (key) trigger points often elicit satellite (remote) trigger points.75 Studies that have evaluated trigger points with electromyography have found both increased (amplitude and duration)76 and decreased77 myoelectric activity, with the decreased activity being attributed to muscle weakness. Irritable muscles can lead to stiffness, spasms, and muscle fiber shortening, which impedes relaxation and reduces strength.78

Although ultrasound and other tests have been advocated to identify trigger points and myofascial neck pain, physical examination remains the reference standard.79 One meta-analysis performed in people with chronic neck pain found that the point prevalence of active trigger points ranged from 14.8% in the right levator scapula to 38.5% in the right upper trapezius.80

The treatment of myofascial pain should be multimodal and should include correcting underlying structural and postural imbalances, physical therapy (for example, massage and range of motion exercises), drugs, and psychotherapy (including cognitive behavioral therapy and biofeedback).81 When conservative options fail, trigger point injections should be considered as adjunctive therapy. This procedure involves inserting a small needle into the taut muscle band(s) in the area of maximal tenderness, which should elicit a local twitch response. A small amount of drug is injected, after which the needle is withdrawn and redirected into the hyperirritable region from different directions. The precise mechanism by which injections inactivate trigger points is unknown, but hypotheses include mechanically disrupting the abnormal taut bands, inhibiting nociceptive feedback loops, increasing local endorphin levels, and releasing cellular potassium, which interferes with nerve conduction.81

Many randomized trials have evaluated trigger point injections, but they generally have methodological flaws such as use as a standalone treatment and the lack of a true placebo group (dry needling could be beneficial).82 Some evidence suggests that trigger point injections alone are not more effective than less invasive treatments such as laser and ultrasound, and that injections with anything, including saline, are better tolerated and more effective than dry needling.838485 One randomized trial in 80 people with chronic whiplash disorders found no significant benefit for dry needling with exercise compared with exercise and sham needling.86 There is little evidence to support one injectate over any other including botulinum toxin,8485 which purportedly works by inhibiting excessive release of acetylcholine, thereby diminishing aberrant muscular contractions.87 An RCT in people with neck and shoulder pain after a whiplash injury found that up to three injections with sterile water significantly reduced pain scores compared with saline three months after treatment (mean decrease in pain score 1.7 v −0.4; P<0.02).88

Cervical facet joint pain

On the basis of medial branch (facet joint nerve) blocks, the prevalence of cervical facet joint pain is estimated at 40-55% in patients with neck pain with or without whiplash.89 In the absence of a history of cervical facetogenic pain or physical examination signs correlating with such pain,90 injections of the medial branches innervating the joints, or the joints themselves, are considered the reference standard for diagnosis.91 However, without a standard for comparison, the accuracy of cervical facet blocks cannot be known, and their utility is controversial. Consequently, the best way to view these blocks may be as “prognostic” blocks before radiofrequency ablation (RFA). Many experts advocate performing placebo controlled or comparative local anesthetic blocks to improve accuracy, because single blocks carry a false positive rate of 27-60%.8992 However, having a prevalence (true positive) rate of 50% coupled with a false positive rate approaching 50% is irreconcilable, leaving little room for true and false negative blocks. Moreover, the prevalence rate in studies does not significantly change with the diagnostic criteria—it is similar regardless of the threshold for a positive block or whether double blocks are used.89 No studies have examined the utility of multiple blocks in the neck, but a comparative effectiveness study done for suspected lumbar facet joint pain found that double blocks resulted in a lower overall success rate and higher cost than proceeding straight to RFA without a screening block.93 The only study that evaluated outcomes of cervical facet RFA in people selected without diagnostic blocks reported 55% and 30% success rates at two months and three years, respectively.94 In one three phase crossover study that compared local anesthetic blocks with lidocaine and bupivacaine versus placebo controlled blocks, comparative blocks were shown to have a specificity of 88% but a sensitivity of only 54%.95 As a screening tool for a relatively safe and effective treatment in which alternative treatments include opioids or surgery, a high sensitivity and negative predictive value are desirable.

The higher prevalence of facet joint pain in the neck than in the low back is attributable to the relatively larger size of the joints compared with the discs and differences in function (less motion in the lower back).91 Given the lower prevalence of facet joint pain in the lumbar spine, a higher false positive rate might be expected in the lumbar spine, but this is not the case.96 This underscores the inherent limitations of using diagnostic blocks to identify a painful structure in the absence of a reference standard. A crossover study evaluating the accuracy of diagnostic intra-articular facet blocks found no significant difference in positive rates between anesthetic, placebo, and sham injections.97

Intra-articular steroid injections are sometimes used to treat cervical facet joint pain, but the technical failure rate is high, the only controlled trial evaluating steroid injections found no benefit,98 and most systematic reviews have concluded that these injections are ineffective.91 Clinical studies have found C2-3 and C5-6 to be the most commonly symptomatic levels,99100 whereas radiological studies have found C3-4, C4-5, and C2-3 to be more commonly affected.101102 Radiological evidence of degeneration is also common in asymptomatic people.103 Along with the facet joints, the atlanto-axial and atlanto-occipital joints can be a source of pain. These joints are not amenable to ablation, and the evidence supporting intra-articular steroids is anecdotal.104

RFA of the medial branches is considered by many to be the standard for treating cervical facet joint pain. However, the evidence supporting RFA is mixed. In the cervical spine, a placebo controlled randomized trial performed in 24 patients with whiplash injury who experienced concordant relief with three controlled diagnostic injections reported significant improvement in the treatment group compared with sham lesioning for a median duration of 263 days (58% of patients were pain free at six months in the treatment group v 8% in the control group; P=0.04).99 The only other controlled trial randomized 12 people with cervicogenic headache to RFA or sham treatment of the C2-6 medial branches.105 Although patients received comparative local anesthetic blocks before treatment, the results were not part of the inclusion criteria. Three months after treatment, four people in the RFA group versus two people in the sham group reported 30% or more improvement in pain, although no differences were noted afterwards. Thus, the evidence supporting RFA for facetogenic pain is weak.

Cervical discogenic pain

Degenerated discs contain high levels of pro-inflammatory mediators.106 More than 70% of people without neck pain have clinically significant disc degeneration by their mid-40s,107 with the prevalence rising above 85% by age 60.108 Disc degeneration also increases the likelihood of herniation.109 No markers can distinguish a painful from a non-painful degenerative disc,106 although there is evidence to support the role of low grade infection in some people.110 The high prevalence of neck pain and disc abnormalities in asymptomatic people provides the conceptual appeal for discography, which is advocated as the only test that connects disease to symptoms.

Provocation discography operates on the premise that increasing intradiscal pressure by contrast injection will provoke concordant symptoms at painful levels, but prevalence studies performed in the neck have yielded disparate findings. Retrospective studies (n=173 and n=31) 111112 reported at least one positive level that provoked concordant symptomatology in 86% and 84% of patients, respectively. In contrast to the lumbar spine, a negative control disc was not required for a positive discogram. Another retrospective study evaluating a battery of diagnostic injections in 143 people with chronic neck pain reported a 16% prevalence rate of discogenic pain.113 A similar prevalence study found that 41% of patients with chronic neck pain had both positive discography and positive facet blocks, whereas 20% had only positive discography.114 Along with differences in technique, another reason for the discrepancies relates to the clinical acumen of the referral source. For example, an experienced surgeon who selects discography candidates carefully will have a higher positive rate than a less experienced practitioner. A systematic review evaluating the accuracy of cervical discography found prevalence rates of 16-53%.115

Discography is an invasive procedure that carries a small risk of catastrophic consequences and a high false positive rate in certain populations.116 Consequently, the evidence for using discography as a screening tool to improve surgical selection is limited. Whereas two studies reported improved fusion outcomes and a lower incidence of postsurgical adjacent segment disease when discography was used to select patients for surgery, these studies were retrospective and conducted before radiological imaging was routine.117118

High quality studies of treatments for cervical discogenic pain without radicular symptoms are lacking, and surgery is a mainstay of treatment in some circles. In general, surgical treatment for common degenerative conditions is widely acknowledged to be inferior to the treatment for mechanical pain accompanied by neurological symptoms, including in the neck. As noted in the surgical section, cervical disc arthroplasty and fusion are the two main surgical treatments for degenerative spondylosis, although in clinical trials and practice these are generally used in people with radicular or myelopathic symptoms (see surgical section for evidence). One review found no evidence to support the use of cervical disc arthroplasty and fusion in the absence of radiculopathy, although cervical spine fusion is often used to treat instability or spinal deformities such as cervical kyphosis in people with neck pain who present without neuropathic symptoms.119 The evidence supporting other treatments for discogenic pain, including epidural and intradiscal injections, and thermal ablation, is weak and inconsistent.120121122123124

Neuropathic pain

Cervical disc herniation

The annual incidence of cervical radiculopathy resulting from disc protrusion or degenerative spondylosis (or both) is estimated at 1-3.5 per 1000 person years, peaking in the sixth decade of life.18125126 A population based study reported a 0.055% incidence of radicular pain secondary to a cervical herniated disc, although estimates based on patients seeking medical attention underestimate incidence.127 Studies in asymptomatic volunteers report the prevalence of disc herniation as 2-23%, with a median of 11%. Most studies reported no significant differences between the sexes.108128129130

Unlike the lumbar spine, where the traversing nerve root is most commonly irritated, in the cervical spine disc herniations and spondylosis most often affect the exiting nerve root, so a C6-7 disc herniation will usually cause C7 symptoms. The most commonly affected levels are C7 (45-60%), C6 (20-25%), and C5 and C8 (10%).318 Not all radicular symptoms result from mechanical nerve root compression. Similar to the lumbar spine, cytokines and other inflammatory mediators play a pivotal role in cervical radicular pain.131132

Few non-surgical treatments have been studied for cervical radiculopathy. Two RCTs evaluating ESIs yielded mixed results,133134 with the study that used a transforaminal approach reporting a negative outcome.134 Most systematic and evidence based reviews have concluded that transforaminal ESI provides more benefit than interlaminar injections, but its use in the neck is limited because of the risk of catastrophic complications such as spinal cord infarction, particularly with depo-steroids.124135 Comparative effectiveness studies by one group showed similar benefit in pain relief, function, and other outcome measures such as opioid consumption and employment for both steroids and local anesthetic in a variety of cervical diseases including disc hernation.120136137138 Another randomized study in 169 patients with radicular pain found the combination of ESI and conservative treatment consisting of physical therapy and the adjuvants nortriptyline or gabapentin (or both) provided superior relief compared with either treatment alone (mean reduction in pain score of 3.1 in the combination group versus 1.9 in the others at one month; P=0.035).139 Together, these studies provide limited evidence that ESI is more effective than epidural non-steroid injections but moderate evidence that a series of epidural injections can provide sustained relief. One criticism about ESI studies that use epidural injections as a control group is that epidural non-steroids are not a true placebo.140 The only placebo controlled study to evaluate pulsed radiofrequency of the dorsal root ganglia reported benefit at three months in 23 patients with chronic cervical radicular pain (82% v 33% experienced positive outcome at three months; P=0.02).141

The results of controlled trials evaluating first line neuropathic drugs for radicular pain have been disappointing. Apart from the aforementioned study that found only small differences between adjuvants and ESI,139 no randomized studies have evaluated drugs solely for cervical radicular pain. Studies evaluating antidepressants and gabapentinoids for lumbosacral radiculopathy have yielded mixed results,142143144145146147 with the best studies being negative.143144147 A placebo controlled trial evaluating pregabalin in 19 people with cervical and lumbar radicular pain found no benefit for pregabalin.146

Cervical spinal stenosis

Spinal stenosis can be classified as central, involving the lateral recesses, or foraminal, with the last two types generally affecting the exiting nerve root(s). The incidence of symptomatic central cervical spinal stenosis, in which the spinal canal diameter measures less than 10 mm, is estimated to be one in 100 000 but significantly increases over the age of 50 years.148 Spinal stenosis has many causes, which can broadly be categorized as congenital (for example, short pedicles), spondylotic (for example, degenerative discs, hypertrophied facet joints and ligaments, and osteophytes), iatrogenic (for example, surgery), traumatic, metabolic (for example, Paget’s disease), and rheumatologic (for example, spondyloarthropathy). In adults, degenerative spondylosis is by far the most common. Risk factors for stenosis include genetics, older age, and possibly occupation—for example, porters, high performance aviators, and athletes such as rugby players are thought to be at increased risk.25 Unlike radicular pain from a herniated disc, spinal stenosis often results in multi-level neuropathic symptoms. When spinal cord compression occurs (for example, myelopathy), symptoms may also include gait and balance disturbances, deterioration in fine motor skills, and incontinence. Hypotheses for the pathogenesis of symptoms include chronic neurogenic compression and ischemia of the spinal cord and nerve roots from venous congestion (fig 7).148149


Fig 7 Axial view of the cervical spine showing potential causes of spinal stenosis. Courtesy of Frank Corl (Mayo Clinic)

Few studies have evaluated epidural steroids for cervical stenosis. Among the randomized trials that assessed cervical ESI, two included patients with stenosis; the first found that one to three cervical ESIs performed with steroid and local anesthetic resulted in superior pain relief and range of motion compared with the same solution injected intramuscularly for up to one year (68% v 11.8% experienced >50% pain relief; P<0.001), a positive study.133 The second found that conservative therapy and a series of ESIs provided significantly greater improvement than either treatment as stand-alone therapy in 169 patients, as noted above.139 Another study comparing a series of ESIs to epidural local anesthetic found that both groups experienced significant, comparable, long term improvement.136 In both the lumbar and cervical regions, it is widely acknowledged that the benefits afforded by ESI for herniated discs are superior to those for spinal stenosis (table 2).124 `

Table 2

Randomized controlled trials evaluating procedural interventions for neck pain*

View this table:

Exercise and integrative medicine

Exercise is often touted as an effective treatment for chronic pain, although a recent Cochrane review found that the quality of evidence is low.158 Exercise may be beneficial for neck pain by stimulating endorphin secretion, improving sleep and mood, and reversing or preventing deconditioning. Exercise programs have been shown to prevent spinal pain in general,35 but few studies have been performed in neck pain alone, and these have yielded mixed results.159160

The effects of exercise on whiplash injuries have also been mixed, with one randomized trial comparing six weeks of exercise in addition to advice with advice alone (n=134),161 and another comparing a 10 week multimodal physical therapy program with self management (n=71).162 The first found small to moderate effect sizes in the short term (immediately after treatment) but no long term benefit.161 The second study found that people allocated to receive the multimodal program reported significantly better pain relief and functional capacity, again assessed immediately after the treatment period (effect size 0.48; P=0.04).160 A larger more recent randomized study in patients with chronic whiplash found no difference between advice and a physical therapy exercise program up to 12 months of follow-up (table 3).176

Table 3

Summary of systematic reviews and meta-analyses of exercise and integrative medicine treatments for neck pain*

View this table:

Integrative medicine, also known as complementary and alternative medicine, encompasses a broad range of treatments including acupuncture, spinal manipulation, massage, meditation, and yoga. Integrative treatments can be delivered by licensed practitioners (such as acupuncturists), physicians, or by self care (for example, meditation). In addition to treatments that require specialized training (such as Tai Chi and yoga), herbal therapies such as soy, turmeric (curcumin), polyphenols, and omega 3 fatty acids have been shown in clinical studies to exert analgesic effects, although none has been specifically studied for cervical pain.177 Despite the fact that integrative therapies are often not reimbursed by insurance companies, population based surveys suggest that the use of such therapies exceeds 50% in some countries, being slightly higher in Asia and Australia. Moreover, people often use several different integrative treatments concurrently.178179 Consequently, clinicians should be aware of the effectiveness of the integrative medicine treatments that are commonly used for neck pain and need to understand which patients might benefit. Table 3 summarizes studies of integrative medicine for neck pain.


A broad range of surgical techniques are used to treat neck pain, with the indications for surgery being dependent on the underlying source of pain. For cervical radiculopathy, a randomized trial that compared anterior cervical decompression and fusion plus physiotherapy with physiotherapy alone (n=59) showed that patients allocated to surgery experienced significantly greater reductions in neck pain (39 mm v 19 mm; P=0.01) and neck disability (21% v 11%; P=0.03) compared with the non-operative group at five to eight year follow-up.180 However, no significant differences were seen for arm pain or self assessed health status. A three arm randomized study compared anterior decompression and fusion surgery, physical therapy, and cervical collar immobilization in 81 patients with cervical radiculopathy. It found significantly greater short term reductions in pain and neurological dysfunction in the surgical group compared with the collar group and non-significantly greater decreases in pain compared with the physical therapy group (present mean pain reduction in surgical group −10 v −9 in the therapy group and −1 in the immobilization group; P<0.01 v the immobilization group), although no significant differences were seen at 15 months.181

Various surgical options are available for the treatment of cervical spondylotic myelopathy including corpectomy and fusion, anterior cervical discectomy and fusion, laminoplasty, and laminectomy and fusion.182 However, randomized trials provide limited clinical outcomes to guide treatment. In a small (n=64) randomized study comparing surgery with conservative care for cervical myelopathy, measures of physical and neurological functioning were similar between the two groups at two and 10 year follow-up.183184 In addition, surgical treatment of spondylotic myelopathy is associated with a complication rate of 11-38%.185 Predictors of complications include older age, longer duration of surgery, and two stage surgery.185 Given the natural course of cervical radiculopathy and myelopathy, and the modest predominantly short term benefits of surgery, conservative treatment with surveillance for neurological progression is a reasonable course for both conditions.

Two widely used surgical options for cervical degenerative disc disease are anterior cervical discectomy with fusion (ACDF) and cervical disc arthroplasty (CDA). Few randomized trials have compared either procedure with non-surgical treatment. However, randomized studies performed for back pain suggest that less than 40% of people who undergo spinal fusion or disc arthroplasty for mechanical pain associated with common degenerative conditions can expect meaningful pain relief or a highly functional outcome that lasts for two years, and that benefits diminish over time.186187 Given the anatomical, functional, and technical differences between surgery for low back and neck pain, it is unclear how generalizable these findings are. In a small (n=47) randomized trial that compared CDA and ACDF, improvements in neck and arm pain were similar for both groups at seven and 10 year follow-up, but patients in the CDA group experienced statistically significantly greater reductions in neck disability scores (8% v 16%) and reoperation rates were lower (9% v 32%).188

Although outcome data from randomized trials are available, the potential benefits of CDA compared with ACDF are still unclear. However, recent systematic reviews suggest that CDA is associated with greater reductions in neck disability, greater satisfaction, fewer complications, lower reoperation rates, and lower rates of adjacent segment degeneration than ACDF.189190 Systematic reviews also suggest that multilevel CDA is as effective as single level CDA and may be associated with greater preservation of cervical motion than ACDF.191192

Emerging treatments

Biological therapies including stem cell therapy, nerve growth factor inhibitors, and platelet rich plasma have been evaluated in other chronic pain conditions and have yielded mixed results. These treatments have yet to be critically studied for neck pain. Future studies should assess their utility for both degenerative and neuropathic conditions.

The use of ketamine to treat refractory chronic pain has generated enormous interest. Ketamine acts as an antagonist at the N-methyl-D-aspartate receptor and is purported to act through the reversal of central sensitization. Controlled trials evaluating ketamine have generally shown short term benefit for neuropathic pain conditions and disorders characterized by central sensitization (such as fibromyalgia) but this drug has not been studied for neck pain.193


In 2013, the Spine Intervention Society updated its guidelines on cervical facet RFA.194 Similar to previous guidelines,195 they advocate performing double blocks before ablation to reduce the false positive rate, either with two different local anesthetics or an anesthetic agent and saline. The guidelines also recommended placing large bore electrodes parallel to the target nerves and performing multiple lesions to increase the success rate but do not mandate sensory or motor testing, the last of which is usually used to minimize the chance of accidental spinal nerve ablation.91 One case report described a case of head drop after multilevel cervical radiofrequency ablation in which motor testing was not used that required eventual spinal fusion.196 A previous article by the guidelines’ author acknowledges that performing two diagnostic blocks before treatment may not be cost effective in some countries, such as the US.197

In the past few years, several guidelines have been published on ESI. A multidisciplinary group representing the neuropathic pain special interest group of the International Association for the Study of Pain provided a weak recommendation for ESI to treat cervical and lumbar radiculopathy, with most of the evidence derived from lumbar studies.198 In a multispecialty working group panel convened under the US Food and Drug Administration’s (FDA) Safe Use Initiative, experts from different organizations provided recommendation on the performance of cervical ESI, which included always reviewing radiological studies, using image guidance with contrast for needle placement, not injecting particulate steroids for transforaminal injections, and not injecting above C6-7. These recommendations were developed because more than 50 cases of paraplegia or death have been reported after cervical transforaminal ESI was performed with depo-steroids. This resulted in the FDA convening a panel in 2014 to evaluate the effectiveness and safety of ESI; the panel concluded that for some ESI (such as cervical transforaminal ESI performed with depo-steroids) the risks may outweigh the benefits.199

Comprehensive guidelines published in 2013 by the American Society of Interventional Pain Physicians on the management of spinal pain found limited evidence for cervical discography and good evidence for the diagnostic validity of cervical medial branch blocks, which they recommended performing before discography.200 However, the accuracy of cervical medial branch blocks to identify a painful joint was not based on an independent test but on double blocks performed with different anesthetics. They reported good evidence for cervical ESI for disc herniation and fair evidence for axial neck pain, spinal stenosis, and post-laminectomy syndrome. For the treatment of facet joint pain, they found fair evidence for therapeutic medial branch blocks and RFA and limited evidence for intra-articular injections. However, this contradicts the rationale for performing diagnostic cervical medial branch blocks with two different local anesthetics, based on concordant relief for the duration of the action of the drugs, which is measured in hours not months.

Guidelines by the Motor Accident Commission of Australia in 2008 provide guidance on the assessment, diagnosis, and prognosis of whiplash associated disorders, as well as recommendations for treatment stratified by chronicity.201 For chronic whiplash, recommended treatments included exercise and advice (grade B), cognitive behavioral therapy (grade C), RFA (grade B), and subcutaneous “trigger point” injections with water (grade C). For acute whiplash, exercise (grade A), advice (grade B), and non-opioid drugs (grade B) were among the treatments advocated. Despite the date, these recommendations are still appropriate.


Neck pain imposes an enormous personal and socioeconomic burden on society, with a prevalence approaching that of low back pain and disability rates ranking within the top five in the US. Yet only a fraction of the resources and attention devoted to low back pain have been dedicated to neck pain. Most cases of acute neck pain, regardless of whether or not they are radicular in nature, will resolve within three months, although a substantial proportion of people will continue to experience low grade symptoms or frequent recurrences. It is crucially important to categorize neck pain as either neuropathic or non-neuropathic because this information is needed to guide investigations (necessity for imaging) and treatment decisions. The facet joints have been implicated as the primary pain generator in nearly half of people with whiplash injuries, but there is a poor correlation between imaging and symptom severity. Topical NSAIDs can be beneficial in people with non-specific neck pain and muscle relaxants are a reasonable treatment choice for acute non-radicular pain. In patients with cervical radiculopathy there is weak evidence to support the use of ESI, and in patients with mechanical pain who respond to diagnostic blocks there is weak evidence to support the use of RFA. Similar to back pain, surgical decompression for patients with radiculopathy can provide short term benefit compared with non-surgical treatment, but the benefits diminish with time. In addition to controlled and comparative effectiveness studies evaluating various treatment options for neck pain, future research should endeavor to determine whether chronic pain can be prevented after an acute pain episode and establish registries to determine large scale treatment outcomes.

Questions for future research

  • How can we better identify people at high risk of developing acute neck pain and those with acute pain who will to transition to chronic neck pain?

  • Can pre-emptive measures (such as ergonomic modification and exercise) and preventive measures (such as advice and mass media campaigns, physical therapies, and drugs) prevent acute pain and prevent pain from becoming chronic, respectively?

  • How effective will registries be in determining effectiveness in large scale populations, identifying phenotypes likely to respond to treatment (precision medicine), and establishing differences in treatment responses between healthcare providers (identifying individuals who select inappropriate candidates for treatment)?

  • Will the growing emphasis on comparative effectiveness and cost effectiveness studies enable providers and third party payers to make long term economically sustainable treatment decisions that improve patient care?

  • What is the role of biological therapies such as cytokine inhibitors, nerve growth factor inhibitors, and stem cell injections in the treatment of neuropathic and degenerative spinal pain?

How patients were involved in this article

The article was sent for review to Geoffrey Chesbrough, who has a long history of neck pain and occipital headaches that developed over the course of his military career. Although no changes were suggested, he emphasized the importance of an accurate diagnosis and the need for doctors to try to tailor treatments to individual patients. He previously underwent a “shotgun” approach, which included epidural steroid injections, cervical facet blocks and radiofrequency ablation, occipital nerve blocks, and trigger point injections, before finally obtaining good, albeit short lasting (less than three months) relief from atlanto-axial joint blocks, which he continues to receive as needed.


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

  • Contributors: SPC conceived the design, wrote and reviewed the article and tables, helped with the figures and is guarantor. WMH wrote and reviewed the article and tables and helped with the figures.

  • Funding: Funded in part by the Centers for Rehabilitation Sciences Research, Uniformed Services University of the Health Sciences.

  • Competing interests: Both authors have read and understood BMJ policy on declaration of interests and declare that they have none.

  • The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.


View Abstract

Log in

Log in through your institution


* For online subscription