OUP user menu

Botulinum Neurotoxin-A for Treatment of Refractory Neck Pain: A Randomized, Double-Blind Study

Diana Miller MD, Diana Richardson MD, Mahmood Eisa MD, Rizma Jalees Bajwa MD, Bahman Jabbari MD
DOI: http://dx.doi.org/10.1111/j.1526-4637.2009.00658.x 1012-1017 First published online: 1 September 2009


Objective. To investigate the efficacy and tolerability of Botulinum neurotoxin-A (BoNT-A) in the patients with refractory neck pain.

Background. An analgesic effect is suggested for BoNT-A by a number of animal studies. Two blinded studies suggested efficacy of BoNT-A in a chronic neck pain.

Methods. Forty-seven subjects were enrolled in a prospective, double-blind, placebo-controlled study. A total of 150 to 300 units of BoNT-A were injected into the neck and shoulder muscles based on pain localization. Subjects completed the visual analog scale (VAS), Pain Frequency Questionnaire and the Modified Oswestry Pain Questionnaire (MOPQ) at baseline, 3 and 8 weeks after the treatment. The primary outcomes consisted of: 1) ≥50% improvement on the VAS; and 2) ≥30% reduction in pain day frequency. The secondary outcome was an improvement of ADL in MOPQ. Excellent responders (ERs) were those who met all three outcomes.

Results. At 2 months, a significant reduction in the mean VAS (pain intensity) was noted in the BoNT-A group compared with the placebo (P = 0.0018, CI 95% from 2.51 to 7.89). At 2 months, there were six ERs in the BoNT-A group and one ER in the placebo group (P = 0.0152).

Conclusion. Administration of BoNT-A into the neck and shoulder muscles for treatment of chronic refractory neck pain met one of the two primary outcomes: reduction in pain intensity. More ERs were noted in the Botox group.

  • Botulinum Neurotoxin-A
  • Neck
  • Persistent Pain
  • Placebo
  • Outcome Measures


The National Health Interview Survey of 2002 reported that 13.8% of the US population (28,401,000 adults) suffer from chronic neck pain [1]. It is estimated that health care costs for the treatment of neck pain ranges from $3,300 to $6,000 per person annually [2].

The treatment of chronic neck pain includes oral administration of non-steroidal anti-inflammatory drugs, muscle relaxants, anti-epileptic, and opioid analgesics. Refractory cases may require intravenous or epidural administration of steroids or anesthetic agents and/or surgery [3]. Up to 40% fail to improve after surgical intervention [4].

An analgesic effect for Botulinum neurotoxin-A (BoNT-A) has been reported in animal studies [5–10]. Human clinical trials also support an analgesic effect for BoNTs [11–16]. We investigated the effects of BoNT-A (Botox-Allergan Inc., Irvine, CA) on refractory neck pain in a prospective, double-blind, placebo-controlled study.



Following the approval from the Human Investigation Committee, the subjects were recruited from Yale-New Haven hospital clinics and through advertisement. Upon meeting the inclusion and exclusions criteria (Table 1), they were enrolled to participate in the study. All subjects signed an informed consent.

View this table:
Table 1

Inclusion and exclusion criteria

Inclusion Criteria
  • age 18 years or older
  • both sexes, all races and ethic groups
  • VAS score > 4 and/or pain frequency > 5/month
  • Refractory neck pain; failed conventional treatment
  • Recent MRI showing no acute neck pathology
Exclusion Criteria
  • Age below 18
  • Diseases of neuromuscular junction (myasthenia gravis, etc.)
  • Drugs impairing neuromuscular transmission
  • Diagnosis of cervical dystonia
  • Previous treatment with BoNTs for spasticity, headache, cosmetic etc.

Randomization, Blinding, Rating Scales

Randomization, blinding, and drug preparation was performed by the Yale Research Pharmacy using a computer software. Each subject identified their areas of pain by marking an anatomical neck diagram. Pain was characterized by severity using the Visual analog scale (VAS). Pain frequency was tallied as the number of pain days within a 4-week interval. Impact on activities of daily living (ADLs) was rated using the Modified Oswestry Pain Questionnaire (MOPQ). All scales were completed at baseline, 3 weeks and 8 weeks after treatment. The monitoring intervals were chosen to capture peak treatment effects based on clinical practice observations. The VAS is a straight line from 0–100 mm, denoting pain intensity, where zero means no pain and 100 means the most severe pain ever experienced. Subject rated their satisfaction with treatment as very satisfied, somewhat satisfied, neutral, somewhat dissatisfied, or very dissatisfied. The Oswestry pain questionnaire is a validated questionnaire for low back pain that reflects performance of ADLs (sitting, standing, etc.) via 10 questions (graded 0–4). The MOPQ used for this study included two additional questions pertaining to driving and watching television. The rating and data collection was performed by a health care provider different from the injecting physician.


The subjects were treated with either the study drug BoNT-A or placebo, based on location and severity of symptoms. For unilateral neck pain, subjects received injections into the upper, middle, and lower cervical muscles (splenius capitis, and cervicis), and into the ipsilateral shoulder trapezius muscle (Figure 1a). Additional injections were given into the upper back and peri-scapular muscles (rhomboids and trapezius) in subjects with an extended pain distribution (Figure 1b). Subjects with bilateral neck pain received bilateral injections using the same technique. Each subject received only one set of injections intramuscularly. The dose per injection site was 20 units (0.2 cc or equivalent saline), and the total dosage varied from 150 to 300 units (or 1 to 3 cc of saline). The total dose per session was determined at the discretion of the injector based on symptom location and extent of pain, based on the patient's completed anatomical diagram. For instance, some patients with extensive unilateral pain received doses over 150 units (Figure 1b) Electromyography was not performed.

Figure 1

(a) Unilateral neck pain confined to the neck and upper shoulder. Points of injections (trapezius, splenius). (b) Unilateral neck pain extending to peri-scapular region. Points of injections (additional rhomboid). In patients with bilateral neck pain (19 out of 47) injections were bilateral using the same scheme.

Statistical Analysis

The primary outcomes in this study were improvement on the VAS and in the pain frequency questionnaire. Improved MOPQ was the secondary outcome. A paired t-test (P value: <0.05, confidence interval 95%) was used to measure the significance of response to the studied drug.

In addition, we defined a group of subjects as excellent responders (ERs) when the following criteria were met:

  1. Showed ≥50% improvement in VAS.

  2. Demonstrated ≥30% decrease in the number of pain days (over 4 weeks).

  3. Showed two grades or more improvement in at least two subsets of MOPQ.

The excellent responders in both groups were statistically evaluated by Fisher exact test (two-tailed).


A total of 47 subjects were enrolled in the study (Table 2; clinical data). The group consisted of 18 men and 29 women with mean age of 52.6 years (SD = 14.32). Pain duration ranged from 6 months to 40 years (mean 9.1 years). Thirty-three subjects had a history of neck injury (motor vehicle accidents 36%, falls 8.5%, work-related injury 4.3%, assaults 2%).

View this table:
Table 2

Clinical data show total number of patients who enrolled in the trial, the number in each group, the duration of follow-up, and the number that completed the study. A total of 55 patients were screened and 8 seemed to be ineligible

Registered eligible patients (N = 47)
Not randomized (N = 0)
Randomized (N = 47)
Received Botulinum Toxin A (N = 25)Received Placebo (N = 22)
Follow up (N = 23)Follow up (N = 22)
3 weeks, 2 months3 weeks, 2 months
Withdrawn (N = 2)Withdrawn (N = 1)
Lost to follow up (none)Lost to follow up (none)
Completed trial (N = 23)Completed trial (N = 21)

All patients failed multiple medications for pain control. These included non-steroidal analgesics, muscle relaxants, and opioid analgesics. Twenty of 47 patients (44%) have been or were still taking opioid analgesics. Six had epidural steroid injections and seven had surgical treatment. Patients' analgesic medications were not changed during the study. There were no cases of workman compensation and litigation in this study. Table 3 contains details of subject's demographic data and medications.

View this table:
Table 3

Demographic data of 47 randomized patients

PatientsSalineBotulinum Toxin A
Sample size2225
M/F 6/1612/13
Mean age (years)51.1 (27–80)52.12 (31–83)
Mean pain duration (years) 6.9 (0.5–26)10.59 (0.75–40)
Patients with neck surgery 5 2
Patients with neurological deficits 9 7
Trauma neck related1518
Pain unilateral/bilateral15/712/12
Root pain1719

The groups were comparable except for the duration of neck pain, which was, on average, 3.5 years longer in the BoNT-A group (responders and non-responders) and female predominance in the placebo group (Table 3).

On ERs, the mean duration of pain was 14.5 years, SD 8.43. The mean dose of BoNT-A was 188.33 units and SD 57.16 for ERs, versus 210.55 units, SD 45.95 for the rest of the Botox group. The clinical diagnosis in majority of the patients was neck pain associated with degenerative cervical spine disease. In 45 of the 47 patients, the anatomical diagnosis was confirmed by MRI, which showed cervical spondylosis, canal stenosis, foraminal encroachments, and chronic disc protrusions. None of the subjects had cord compression or acute pathology. The MRIs were normal in two subjects. Sixteen patients had subtle neurological deficits consisting of dermatomal sensory loss [6], mild muscle weakness [5], and reflex asymmetries [8].

A total of 44 subjects completed the study (Table 2; clinical data). Two patients withdrew after signing the informed consent. One withdrew a couple of days after being injected, stating that immediate relief did not happen as he expected. Of the 44 patients, 23 received Botox and 21 received placebo. The improvement of mean VAS scores for the subjects in the BoNT-A group was significant compared with placebo group at 2 months (P = 0.0018, CI 95 % from 2.51 to 7.89) (Figure 2). Pain frequency and MOPQ values showed no significant difference between groups.

Figure 2

Pain intensity measured by VAS comparing the means between the Botox group and the placebo group. Botox versus saline: 3 weeks: ns, 2 months: P < 0.0018.

At three weeks both groups had six ERs (P = ns). At 8 weeks, the number of ERs in the Botox group remained unchanged, whereas only one ER remained in the placebo group (P = 0.0152, two-tailed Fisher exact).

All patients tolerated the procedure well. Two subjects (one in the BoNT-A group and one in the placebo group) reported a transient flu-like reaction that lasted for 3–4 days. One female subject with a small neck (BoNT-A group) reported mild “neck weakness” that lasted 4 weeks.


The data from animal studies suggest an analgesic action for BoNTs. In rats, the injection of formalin into the paw causes severe pain associated with local inflammation and tissue accumulation of glutamate. Pre-treatment of the paw with intramuscular administration of BoNT-A reduces the pain significantly, along with reduction of tissue inflammation and glutamate accumulation [5]. BoNT-A inhibits the release of pain neurotransmitter substance P in vivo [6] and reduces the discharge of sympathetic neurons, with the potential to reduce sympathetically maintained pain [7]. Intramuscular injection of BoNT-A diminishes the discharge of muscle spindles [8]. This reduces the sensitivity of the spinal cord's wide range of action sensory neurons, which, when sensitized (chronic pain conditions), perceive non-nociceptive pain stimuli (such as those coming from muscle spindles) as nociceptive. In rats, thermal and mechanical hyperalgesia caused by partial sciatic nerve transaction showed improvement with local BoNT-A treatment [9]. BoNT-A or its metabolites may influence the spinal cord function via retrograde transmission to ipsilateral hemi-cord after intra-sciatic injection [10].

In humans, the analgesic effect of BoNTs is well known through the experience of treating cervical dystonia. In cervical dystonia, BoNTs (A or B) significantly relieves the neck pain in 60–80% of the patients. This analgesic effect appears to be independent from and precedes the improvement of neck posture. [11,12]. Other prospective and placebo-controlled studies reported pain relief after administration of BoNT-A in subjects with plantar fasciitis [13], pelvic pain [14], migraine [15], and low back pain [16]. More recently, intradermal injection of BoNT-A in human subjects relieved the neuropathic pain and allodynia [17], as well as the pain associated with diabetic peripheral neuropathy [18].

Prospective studies of BoNT treatment in chronic neck pain are few. The short-term double-blind, placebo-controlled studies of Freund and Schwartz (whiplash injury) [19] and Ojala et al. [20], both with BoNT-A, depicted significant pain relief at 4 weeks (P = 0.008). The last study additionally showed a trend toward improvement in global quality of life (P = 0.097). In another double-blind, placebo-controlled study [21], injection of BoNT-A (Dysport) into trigger points resulted in significant pain relief in patients with myofascial pain syndrome affecting neck and shoulder muscles. Vasan et al. [22], in a prospective open-label study of 16 subjects with neck dissection surgery, reported relief from chronic and shooting pain (P = 0.005) after injection of 80 to 320 units of BoNT-A (Dysport) into the neck and shoulder muscles. Conversely, Wheeler et al. [23] reported no significant difference between Botox and placebo in a double-blinded study of 50 subjects with chronic neck pain. In this study, however, 17 of the 25 treated subjects had BoNT-A injected into the trapezius muscle only. The remaining eight had injections into seven neck sites, which did not include the middle and upper cervical regions. In a different study, however, these authors found that after a second injection, considerably more patients were pain-free in the BoNT-A group compared with the placebo group [24].

Our study met one of the two primary outcomes: i.e., improvement of pain intensity. Furthermore, 26% of subjects (6 out of 23) in the BoNT-A group reported improvement in all three measures (the intensity, the frequency, and the ADL = excellent responders) at 3 and 8 weeks.

The subjects' mean pain intensity showed a steady trend toward improvement in the BoNT-A group reaching statistical significance only at 8 weeks (Figure 2). It is not clear why statistical significance was not achieved at 3 weeks. Possibilities include an early large placebo effect that vanished by 8 weeks. Another possibility is that the chronic nature and complexity of pain in this cohort prevented an earlier response after a first time treatment.

Our study has limitations: First, the study group is small hence the study of a larger group of patients with refractory neck pain may produce different results. Second, the study did not provide follow up beyond 2 months. A longer-term outcome data, preferably with repeated injections, can provide more valuable information. Despite these limitations, the data from our study, combined with the information from recent literature, encourages future investigations of the efficacy and safety of BoNT-A in chronic neck pain. Recommendation for further studies should include a larger study group, comparable study subgroups, using more precise injection techniques, and rigorous rating scales for measurement of the quality of life and activities of daily living.


  • This study was supported by a grant from AllerganInc., Irvine, CA.

  • Disclosure: Bahman Jabbari, MD, received research and educational grants from Allergan, Solstice, and UCB companies.


View Abstract