British Journal of Rheumatology 1998;37:1118–1122

CHRONIC NECK PAIN: A COMPARISON OF ACUPUNCTURE TREATMENT AND

PHYSIOTHERAPY
J. DAVID, S. MODI, A. A. ALUKO, C. ROBERTSHAW and J. FAREBROTHER
Royal Berkshire and Battle Hospitals NHS Trust and University of Reading, Reading

SUMMARY
Objective. To evaluate the effectiveness of acupuncture, as compared with physiotherapy, in the management of chronic
neck pain.
Design. Seventy adult patients with non-inflammatory neck pain of >6 weeks duration and with no abnormal neurology
were randomly assigned to receive either of the treatments. Thirty-five patients were included in each group.
Outcome measures. Pain by visual analogue scale and neck pain questionnaire, improvement in range of movement of neck
relative to baseline, and well-being (general health questionnaire). Measurements were recorded at the start of treatment, at 6
weeks and at 6 months.
Results. Both treatment groups improved in all criteria. Acupuncture was slightly more effective in patients who had higher
baseline pain scores.
Conclusions. Both acupuncture and physiotherapy are effective forms of treatment. Since an untreated control group was
not part of the study design, the magnitude of this improvement cannot be quantified.
K : Neck pain, Acupuncture, Physiotherapy, Chronic pain.

N pain is common. Thirty per cent of the working as chiropractic, osteopathy, acupuncture and aroma-
population aged 25–29 yr and 50% over 45 yr report therapy, have been cited as potentially useful.
one or more attacks of stiff neck [1]. Neck pain exists This study sets out to evaluate the role of acupunc-
in all occupational groups. In some industries, neck- ture and physiotherapy in the management of chronic
related disorders may account for as many days of neck pain by randomly allocating matched patient
absenteeism as low back pain [2]. Neck pain has groups to these two modalities of treatment.
significant morbidity, such that it comprises a signific-
ant number of referrals to the rheumatologist [3]. METHODS
Symptoms are most prominent, whilst signs may be Subjects
limited to neck tenderness or a reduced range of neck All patients aged 18–75 yr, who were referred over
movement. Objective neurological abnormality is rare. a 1 yr period by general practitioners, orthopaedic
Therapy is aimed at relief of pain and stiffness. surgeons and rheumatologists with neck pain of >6
Many treatments are accepted as standard forms of weeks duration, were suitable for the study. The type
practice. These include non-steroidal anti-inflam- of neck pain included postural neck pain, chronic
matory drugs and physical measures such as heat, whiplash injury with no instability on cervical spine
ultrasound and manipulation. There is, however, little X-ray, occupationally related neck pain such as in
good evidence for their accepted use. Education is VDU operators, and neck pain related to cervical
important. spondylosis. Patients who had had previous acupunc-
A physician’s survey about attitudes to treatment ture or physiotherapy for neck pain were excluded.
[4] showed that active exercise, traction, transcutane- Other exclusions included patients with neurological
ous electrical nerve stimulation and ultrasound were signs present due to nerve entrapment, anti-coagulated
perceived to be the best methods for the treatment of patients, primary fibromyalgia syndrome, inflammat-
neck pain. However, a meta-analysis of studies of ory neck pain, e.g. ankylosing spondylitis, rheumatoid
conservative management of mechanical neck pain arthritis, and patients currently receiving osteopathy
shows no evidence for many of the standard or chiropractic.
approaches to neck pain used in health care today [5]. Seventy patients were recruited. The referring doctor
Even for the treatments found to have some early was sent a letter explaining the study. The consent of
evidence of support, such as manual treatments in the general practitioner as well as the patient’s fully
combination with other treatments, conclusions must informed consent were obtained. Randomization
be cautious because of the small number of trials on occurred centrally and the patient had to agree to the
which they are based, and the varying quality of study prior to the randomization. Thirty-five patients
study design. were randomly assigned to physiotherapy treatment
A number of complementary medical practices, such and 35 patients to acupuncture. The equal patient
numbers in each group occurred by chance.
Submitted 29 December 1997; revised version accepted 5 June
1998. Study design
Correspondence to: J. David, Rheumatology Department, Battle The patients had three assessments. These were
Hospital, Reading RG3 1AG. performed within 1 week prior to the start of treatment,

© 1998 British Society for Rheumatology

1118
 DAVID ET AL.: ACUPUNCTURE AND PHYSIOTHERAPY FOR NECK PAIN 1119

after the sixth treatment and at the 6 month follow- RESULTS

up. Each assessment was performed without knowing Of the 35 patients assigned to the physiotherapy
the results of the previous assessment. These included treatment, seven did not attend at all, compared to
a visual analogue scale of pain ( VAS ), the Northwick only two patients assigned to acupuncture. Sixty-one
Park neck pain questionnaire (NPQ) [6 ] and a compos- patients had baseline measurements taken and began
ite functional assessment of joint range (neck range of the treatment. Three patients on acupuncture and two
movement). The latter was assessed by having the on physiotherapy failed to attend the 6 week assess-
patient sit on a chair (standardized positioning) with ment. A further one acupuncture patient and four
hips, knees and ankles at right angles, and arms folded physiotherapy patients were absent at the 6 month
across the chest to minimize thoracic movement. The assessment.
back was comfortably supported against the back rest.
Using a Myrin goniometer, the patients were asked to Baseline characteristics
perform flexion, extension, side flexion and rotation of There were 23 females and 10 males assigned to
the head to both left and right. Each movement was acupuncture, and 18 females and 10 males assigned to
performed three times and a maximum joint range physiotherapy. A x2 test indicated no imbalance. The
recorded. These were then scored and averaged. A mean age of the acupuncture group was 48 yr com-
General Health Questionnaire (GHQ 28) was com- pared to 44 yr in the physiotherapy group. Both groups
pleted on all patients at the three assessments [7]. had a similar age range. T-tests and a non-parametric
The physiotherapy was performed by a senior Wilcoxon score test indicated no significant imbalance.
physiotherapist (AAA). An explanation of the therapy The baseline measurements of the various response
was given. Standard localized mobilization techniques measurements were checked in the same way and no
were employed, most commonly Maitland rotation, imbalance was detected. These analyses were repeated
postero-anterior oscillatory movement and longitud- using only those patients with complete data.
inal traction. All patients were given a maximum of The baseline characteristics of the non-attenders
six sessions at weekly intervals. were no different from those of the attenders. Those
The acupuncture was performed by CR and SM. who did not attend after the first session were not
Both are general practitioners and registered with the faring better or worse than the other patients. It is,
British Medical Acupuncture Society. An explanation therefore, reasonable to assume that the response to
of the acupuncture technique was given. Sterile, dispos- the treatment in itself does not affect attendance.
able, 0.25 × 2.5 Acumedic needles were used. The
patient had local needling of trigger points. In addition, Statistics
regional needling (GB21—supraspinatus tendon area) Overall, the majority of patients improved over the
and distal needling (L14—web space between thumb first 6 weeks on both treatments, as seen in Figs 1–3
and first finger) were used. The needles were left in situ and Table I.
for 15 min and manually manipulated once at 7 min.
Electro-acupuncture and moxibustion were not used. VAS of pain
Each patient had six acupuncture treatments at weekly Figure 1 shows the mean VAS scores for patients in
intervals. No side-effects from the acupuncture treat- each treatment group with approximate 95% confid-
ment occurred. ence intervals. The means are very similar at baseline,
are lower at the first assessment and rise slightly at the
end. When the ANCOVA was performed on the trans-
Statistics
The VAS score and NPQ score were percentages,
but treated as continuous variables. An analysis of
covariance (ANCOVA) was performed on each of
these responses at 6 weeks and 6 months, taking into
account the baseline score and other covariates such
as age and sex.
The ANCOVA assumes that the response of interest
is normally distributed. This is not the case with the
variables considered in this study, as they are scores
and therefore all whole numbers. As the range of data
was large, it was felt that this was not too great a
problem.
There was concern that the variability in the scores
was not constant and the analysis was therefore also
performed using a variance-stabilizing transformation.
The results were, however, similar.
The change in GHQ score from baseline and the
joint range scores at 6 weeks and 6 months were F. 1.—Mean VAS scores with approximate 95% confidence
analysed using the non-parametric Wilcoxon rank test. intervals.
1120 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 10

formed scores at 6 weeks, age appeared to have acupuncture. As there were only a small number of
a significant effect and was retained in the model patients in each treatment group, the importance of
(P = 0.06). The baseline VAS score, as expected, was this result is difficult to assess. At the 6 month assess-
a major influence on the score at 6 weeks (P < 0.01). ment, there were no significant effects.
The choice of treatment was not significant (P = 0.18).
Although physiotherapy had lower VAS scores at 6 Neck range of movement score
weeks, the difference between the treatments was not The Wilcoxon test showed a marginally significant
significant. difference between the treatments at 6 weeks
After 6 months, 22 of the 29 acupuncture patients (P = 0.09) with physiotherapy appearing to be slightly
remaining still had lower scores than at baseline. more effective. At 6 months, there was no difference
Fifteen out of the 22 physiotherapy patients remaining between the treatments.
had lower scores and the rest had higher scores. The
ANCOVA showed no significant effects. General Health Questionnaire
Both acupuncture and physiotherapy patients
Neck Pain Questionnaire improved their total GHQ score ( Fig. 3). An
Figure 2 shows the mean NPQ scores at each assess- ANCOVA was performed, but here the residuals
ment for the two treatments. The mean for both looked less normal, because the scores are only out of
treatment groups falls at the 6 week assessment and 28, not 100. There were no significant effects. A non-
then remains fairly constant at the 6 month assessment. parametric Wilcoxon rank test was performed. The
The majority of patients improve on both treatments effect of baseline measurements was allowed for here
( Table I ). At 6 weeks, age, sex and baseline score were by looking at the change from baseline at each assess-
found to be important. Treatment was not significant ment. Neither test was significant (P = 0.50 and 0.71,
(P = 0.72). There was a significant treatment by base- respectively).
line interaction, however. This suggests that individuals The GHQ 28 is subdivided into four sections: A, B,
with low baseline scores may do better on physiother- C and D. Seven questions are asked in each section
apy and those with higher scores may do better on and the patient rates these as more than usual, the

F. 2.—Mean Neck Pain Questionnaire scores with approximate F. 3.—Mean General Health Questionnaire scores with approxi-
95% confidence intervals. mate 95% confidence intervals.

TABLE I
Number of patients in each treatment group improving, staying the same and becoming worse compared to baseline for the different
measurements

6 weeks 6 months

Measurement Treatment Better Same Worse Better Same Worse

VAS score Acupuncture 25 1 4 23 0 7

Physiotherapy 23 0 3 15 0 7
NPQ score Acupuncture 24 1 4 23 2 4
Physiotherapy 27 0 0 19 0 4
GHQ score Acupuncture 21 5 4 20 3 6
Physiotherapy 17 7 1 12 7 3
Joint range score Acupuncture 10 13 7 11 11 7
Physiotherapy 12 11 3 7 8 7
 DAVID ET AL.: ACUPUNCTURE AND PHYSIOTHERAPY FOR NECK PAIN 1121

same as usual, less than usual or much less than usual. significantly more effective than placebo immediately
These were then scored as 0, 0, 1 and 1, thus giving a post-treatment. However, the study has no follow-up
possible total of 28. Section A relates to questions on data. Five of the six patients who had initially received
somatic symptoms, B to anxiety and insomnia, C to mock TNS were later treated with acupuncture and
social dysfunction and D to depression. Again, the improved significantly.
choice of treatment had no effect on the change from In our study, improvement is seen in both the
baseline of the score within each section at 6 weeks acupuncture and the physiotherapy groups post-
and 6 months. treatment, i.e. at 6 weeks. It cannot, however, be
concluded that either treatment is in itself effective as
DISCUSSION this improvement may be due to other factors.
Despite the understanding of the anatomy and pain- However, the improvement was more than one would
sensitive structures of the neck, pharmacological treat- expect from placebo alone. It was felt unethical to give
ment of chronic pain often fails to achieve optimal patients no treatment at all over a 6 month period.
control. Further, potential benefits are often dimin- The natural history of neck pain of the type included
ished by unacceptable side-effects. Physiotherapy is the in this study is, therefore, probably not known.
mainstay of treatment of chronic neck pain. There are, Furthermore, although the reasons for careful stand-
however, very few randomized case-controlled trials to ardization of treatment in this study are obvious, it
support its efficacy. should be noted that, in clinical practice, the therapist
Acupuncture is now seen as complementary to (physiotherapist or acupuncturist) has freedom to
orthodox measures of pain relief. Berry et al. [8] tailor the approach to the individual’s needs. In our
compared the effects of acupuncture, physiotherapy, study, when the two treatment types are compared,
injection plus non-steroidal anti-inflammatory drugs there was no significant effect of the treatment.
and placebo (sugar pill plus sham ultrasound ) on pain However, owing to the sample sizes and the fairly large
and joint movement in 60 patients with shoulder cuff variability in responses, equivalence of the two treat-
lesions. A range of subjective and objective measures ment groups cannot be concluded with full certainty.
were administered before treatment and again at 2 and The only significant effects were found for the NPQ
4 weeks. All patients improved on all measures and no score. Acupuncture appeared to work better at redu-
differences emerged between any of the treatments. In cing neck pain than physiotherapy for patients with
a randomized, controlled study of acupuncture treat- high baseline pain scores.
ment for chronic neck pain, Coan et al. [9] compared A further drawback of the design of this study,
the progress of 15 patients receiving classical acupunc- which compares two treatment groups, is the lack of
ture with a further 15 receiving no treatment. A total blinding. It is clearly not possible to keep the treatment
of 80% of the treatment group showed improvement method secret. However, every attempt was made to
on measures of pain, medication use and activity level. make the assessments as independent and as objective
This was better than the untreated group, where only as possible. Patel et al. [12], in their meta-analysis of
2% showed slight improvement on any measure. randomized, controlled trials of acupuncture in chronic
However, all data were obtained from the patients pain, concluded that whilst potential sources of bias,
themselves and there were no follow-up data. Further, including problems with blindness, precluded a
subjects were recruited through newspaper advertise- conclusive finding, most results apparently favoured
ments and so were self-selected. It is possible that they acupuncture.
were not representative of ‘rheumatological’ type neck CONCLUSION
pain. Their initial attitude to acupuncture may have This study shows that both acupuncture and physio-
been more favourable overall than those of a group therapy appear to be similarly effective in the manage-
presenting at an out-patient clinic. In a comparative ment of neck pain. If the clinician is to choose a single
study of electro-acupuncture at classical sites vs physio- treatment, then perhaps acupuncture might be chosen
therapy in the treatment of neck pain, Loy [10] found if the baseline pain score is high. Furthermore, acu-
67% subjective improvement after 18 thrice-weekly puncture treatment within the NHS costs approxi-
sessions of acupuncture, rising to 87% after 6 weeks mately one-fifth that of physiotherapy per treatment.
of treatment. The comparable figures for physiotherapy The growing magnitude and socio-economic factors
were 31 and 53%, respectively. Similar differences were of neck pain in society demand that more research be
also apparent between the two treatments on objective conducted into the efficacy and effectiveness of treat-
measures of neck movement. The results appeared to ment, and indeed into the methodology of trials of
favour electro-acupuncture, although no statistical physiotherapy and acupuncture. As more studies are
analysis of the data was presented. No follow-up data accumulated, the influences of a number of factors on
were presented either. the results can be explored. Larger trials or meta-
Petrie and Langley [11] assigned 13 patients with analysis using consistent methodology will be required
chronic cervical pain to either classical acupuncture or to determine optimum treatment approaches.
mock TNS. Treatment consisted of eight twice-weekly
sessions of 20 min each and improvement was rated A
by the subjects on a simple five-point scale. Despite The authors wish to thank Mrs Hilary Cook for her
the rather basic rating system, acupuncture proved secretarial expertise, Mrs Judith Harverson for her
1122 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 37 NO. 10

assistance with clinical arrangements and Dr Anthony 6. Leak AM, Cooper J, Dyer S, Williams K et al. The
Bradlow for his helpful comments on the manuscript. Northwick Park Neck Pain Questionnaire, devised to
measure neck pain and disability. Br J Rheumatol
1994;33:469–74.
R 7. Goldberg DP, Hillier VF. A scaled version of the General
1. Cassidy JJ, Hiltner A, Baer E. Hierarchical structure of Health Questionnaire. Psych Med 1979;9:139–45.
the intervertebral disc. Clin Rheumatol 1989;8:282. 8. Berry H, Fernandes L, Bloom B, Clark RJ, Hamilton
2. Kvarnstrom S. Occurrence of musculoskeletal disorders EBD. Clinical study comparing acupuncture, physiother-
apy, injection and oral inflammatory therapy in shoulder
in manufacturing industry with special attention to occu-
cuff lesions. Curr Med Res Opin 1980;7:121–6.
pational shoulder disorders. Scand J Rehabil Med 9. Coan RM, Wong G, Coan PL. The acupuncture treat-
1983;8(suppl.):1–114. ment of neck pain: a randomised controlled study. Am
3. Bland JH. Disorders of the cervical spine. Diagnosis and J Chin Med 1982;9:326–32.
medical management. Philadelphia: WB Saunders, 1987. 10. Loy TT. Treatment of cervical spondylosis. Med J Aust
4. Rush PJ, Shore A. Physicians’ perceptions of the value 1983;2:32–4.
of physical modalities in the treatment of musculoskeletal 11. Petrie JP, Langley GB. Acupuncture in the treatment of
disease. Br J Rheumatol 1994;33:566–8. chronic cervical pain. A pilot study. Clin Exp Rheumatol
5. Aker PD, Gross AR, Goldsmith CH, Peloso P. 1983;1:333–5.
Conservative management of mechanical neck pain: 12. Patel M, Gutzwiller LERF, Paccaud F, Marazzi A. A
systematic overview and meta-analysis. Br Med J meta analysis of acupuncture for chronic pain. Int
1996;313:1291–6. J Epidemiol 1989;18:900–6.