ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.www.acupunctureinmedicine.org.uk/volindex.php 121

Papers

Stephanie L Pradydoctoral studentDepartment of HealthSciencesUniversity of York, UK

Kate Thomasprofessor(CAM research)

Lisa Esmonderesearch officer

School of HealthcareUniversity of Leeds, UK

Simon Crouchsenior research fellow

Hugh MacPhersonsenior research fellow

Department of HealthSciencesUniversity of York, UK

Correspondence:Stephanie L Prady

slp505@york.ac.uk

The natural history of back pain after arandomised controlled trial of acupuncture vsusual care – long term outcomesStephanie L Prady, Kate Thomas, Lisa Esmonde, Simon Crouch, Hugh MacPherson

AbstractIntroduction There is growing evidence about the effectiveness of acupuncture in the short term treatment

of chronic low back pain but little is known about long term outcomes. To address this question we followed

up participants of a past randomised controlled trial of acupuncture to assess outcomes after 5.5 to 7 years.

Methods A postal questionnaire assessing pain, quality of life, disability, experience with back pain and

healthcare resource use was sent to all 239 participants of the York Acupuncture for Back Pain trial.

Results Response to the survey was low at 43.9%. Pain measured by the SF-36 Bodily Pain dimension was

maintained in the acupuncture group since the last follow up 3.5 to 5 years previously. The usual care group

had improved over the intervening years and there was now no evidence of a difference between the groups

(difference -0.4 points, 95% confidence interval -10.1 to 9.7). The results were unchanged on sensitivity

analysis using multiple imputation. In both groups back pain had not completely resolved and worry about

back health was moderate. Physiotherapy and acupuncture were used at similar rates for continuing treatment.

Discussion We theorise that exposure to a short course of acupuncture speeds natural recovery from a back

pain episode, but improvements plateau after two years. Acupuncture is often accessed privately for long term

management of back pain but is rarely available within the health service. While our study methods were robust,

the low response rate means that our findings should be interpreted with caution.

KeywordsBack pain, acupuncture, long term effects, randomised controlled trial.

IntroductionBack pain affects 59% of UK adults at some point in

their lives and results in significant personal and

societal burdens.1 The precise cause of the pain is

often difficult to diagnose and may be related to

ligamentous or muscular strains and sprains,

fibromuscular problems or osteoarthritic processes.2

The natural history of back pain is not well

understood although some data indicate that it is

rarely a self limiting condition.3;4 In 1994 the

estimated cost of low back pain to a UK GP practice

of 10,000 patients was £88,000,5 and it is among the

most costly conditions studied in the UK.6

There are a range of treatment options but

research into the most effective therapy is continuing

and evidence based effectiveness guidelines for

treatment have yet to be implemented in the UK.

Acupuncture is one choice that patients are seeking

out; an estimated 1.6% of the British population

consulted an acupuncturist in 2001 and back

complaints account for more than 12% of

consultations to traditional acupuncture practitioners

in the UK.7;8 Most acupuncture consultations are in

the private sector, with only 14% of primary care

practices offering access within the National Health

Service (NHS) in 2001.9 There is a growing body of

evidence on the effectiveness of acupuncture in the

treatment of chronic low back pain; systematic

reviews have found that acupuncture may be more

effective in the short term relief of pain than sham

treatment or no treatment.10-12 However, these

conclusions are tempered by low study quality and

there are few data on long term (>3 months)

outcomes. To address this gap in the literature, the

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 121

current study sought to examine the long term effects

of acupuncture treatment.

The study was based upon a previous trial

conducted by two of us (KT and HM) and reported

elsewhere.13 Briefly, during 1999 and 2001, 241 adults

who had recently consulted their GP about low back

pain were recruited to a pragmatic trial. Participants

were randomised to either the offer of up to 10

acupuncture treatments, or usual GP care. After 12

months there was a difference of 5.6 points in favour

of acupuncture on the primary measure (SF-36 Bodily

Pain dimension) (95% Confidence Interval, -1.3 to

12.5; P=0.1). At 24 months there was an 8 point

difference (95% CI, 0.7 to 15.3; P=0.03) and at this

time point acupuncture was also found to be cost

effective.14 An intriguing finding from this trial was that

the authors found more evidence of an improvement

between the groups at 24 months than at 12.Apossible

explanation of this is that the positive effects of

acupuncture continue to accumulate over time.

The aim of the current study was to follow up

with the participants of this trial and examine whether

the difference between the groups persisted 5.5 to 7

years after randomisation. We also planned to

examine healthcare resource use and self

management techniques.

MethodsEthical approval

Approval for the project was obtained from York

Local Research Ethics Committee (06/Q1108/31)

and the North Yorkshire Alliance Research and

Development Unit.

Data collection

In July 2006, 239 of the 241 participants in the

original trial were sent a letter informing them of the

new study, together with a participant information

sheet. The two participants who had withdrawn their

informed consent from the original study were not

contacted. Potential participants whose letters were

returned undelivered were contacted by telephone

where possible to establish a new address. Two weeks

after the initial letter was mailed, a questionnaire

together with a participant information sheet was

sent to all addresses that were known or presumed

viable.

Return of the questionnaire included consent for

the questionnaire portion of the study. The contact

details of the study coordinator were made available

for participants to discuss any concerns. If participants

did not return their questionnaire within four weeks

of mailing, they were sent a second packet with a

friendly reminder letter. Finally, non-responders were

contacted by telephone in an attempt to elicit key

outcomes.

Participants who returned a completed

questionnaire or provided outcome data over the

telephone were sent a £5 retail voucher and a copy of

a specially written summary of the trial results. At that

stage participants were informed whether or not they

would be invited for a qualitative interview to talk

about their experiences, which is a parallel study

reported elsewhere.15

Primary outcome measure

In the interest of comparability, the measures and

algorithms used in this follow up were similar to

those used to collect data in the original trial.

The primary clinical outcome measure was the

Bodily Pain dimension of a developmental version of

the Mosby Short Form 36 item non-specific quality

of life questionnaire (SF-36 V1)16 adapted and

validated for the UK population.17 One of the eight

subscales, the Bodily Pain dimension, measures two

items: the amount of bodily pain experienced and

interference of the pain on activities over the previous

four weeks. Higher scores indicate better health

status, and a five point difference in before and after

scores on this dimension is considered the minimum

clinically important change (MCIC) with 10 points

indicating moderate differences.18 The Bodily Pain

dimension has proved sensitive as an overall measure

in back pain.19-21

Secondary outcome measures

The Back Pain Experience Questionnaire is an

unvalidated battery of questions including some

asked previously, such as: presence of back pain;

steps taken to control or prevent pain; worry and

feelings about acupuncture. A numerical rating scale

of the degree of worry felt about back pain was added

for this time point.22

The Oswestry Disability Index (ODI) is a

validated disease-specific measure for back or leg

pain that discriminates change.23;24 An improved 10-

item measure is now validated and recommended;25

however, participants were previously administered

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.122 www.acupunctureinmedicine.org.uk/volindex.php

Papers

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 122

the original nine-item version that was used for this

follow up. The MCIC for the ODI is considered to be

greater than 10 points.26 The remaining seven

dimensions of the SF-36 were scored,18 and data

pertaining to healthcare resource use and work status

were collected.

Statistical procedures

Data were single-entered into SPSS and the primary

outcome data double-entered. Demographic and

clinical outcome profiles were reported using mean

(continuous variables) or percentage (categorical)

differences. Between group differences for all

continuous measures were analysed using linear

regression adjusting for age and baseline score.

Categorical variables from the Back Pain Experience

Questionnaire were dichotomised and analysed using

percentage differences and Fisher’s exact test and

quoted with odds ratios. A question about degree of

worry collected on an ordinal scale was analysed

using Student’s t test. Reports of steps taken to

prevent or minimise the impact of back pain were

analysed using percentage difference.

Data were analysed on complete cases using the

intention-to-treat principle (ITT). Statistical analysis

was carried out using SPSS 14 (SPSS, Inc, Chicago,

Illinois) and R version 2.4.1. Multiple imputation

analyses were carried out using the R libraries ‘mice’

and ‘mitools’. All analyses were two tailed and the αset at 0.05. Ninety-five percent confidence intervals

were calculated for all estimates. Analysis of the

effect of missing data was carried out under the

assumption that missing data were ‘missing at

random’ (MAR) that is, missingness could be

dependent on covariates and clinical outcomes but not

on unobserved outcomes.27 Multiple imputed datasets

were calculated using the method of ‘multiple

imputation by chained equations’ (MICE) using the

default imputation methods provided by the mice

library. Regression results from the multiple

imputations were combined using mitools.

Sample size and study power

Based on pilot data, the original acupuncture trial

was powered to detect a 10 point change (SD 19.3)

on the SF-36 Bodily Pain dimension with α of 0.05

and β-1 of 90%.14 Based on prior attrition of 23.9%

at 24 months, the anticipated response rate for this 5.5

to 7 year follow up was estimated at 50% of the

original study population. The power to detect

differences between groups was retrospectively

calculated from actual response rates using G*Power

for Windows.28

ResultsFlow of participants through the study

At this time point of up to seven years post-

randomisation, we achieved an overall response rate

of 43.9% (105/239) (Figure 1). However, excluding

the 70 people who definitely did not receive the

questionnaire (ie envelopes returned ‘addressee

unknown’ and known deceased), the response rate

was 62.1% (105/169). We feel that this is an

acceptable rate of return given that we did not attempt

to contact or locate respondents in the intervening

years. Similar rates of response were observed in

both randomisation groups. Upon telephone contact,

12 people volunteered the reason why they did not

complete the questionnaire: forgot or had no time

(n=5); pain gone/changed and didn’t think their input

would be relevant (n=2); questionnaire too long

(n=2); done enough for this study already (n=2);

didn’t want to be reminded of current pain (n=1).

Eleven of the 12 volunteered primary outcome data

over the telephone.

Characteristics of respondents and non-

respondents

The characteristics of respondents and non-

respondents are shown in Table 1. Respondents

appeared to be older than non-respondents, and usual

care respondents older than respondents in the

acupuncture group. Respondents may have reported

less pain at the 3-month data collection point than

non-respondents but otherwise characteristics

between groups appeared similar.

Analysis of the primary outcome

All 105 respondents completed the two questions

on the SF-36 that made up the primary outcome. The

primary indication was that both groups had retained

improvements in their pain since randomisation

(Table 2).

The adjusted mean score difference at this time

point was -0.4 points (95% CI, -10.1 to 9.7, P=0.9).

This result should be interpreted with caution as post

hoc calculations indicated the study was only

powered at 64%.

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.www.acupunctureinmedicine.org.uk/volindex.php 123

Papers

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 123

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.124 www.acupunctureinmedicine.org.uk/volindex.php

Papers

Figure 1 Participant flow through study

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 124

Using multiple imputation to correct for possible

dropout based bias and to improve power resulted

in no significant changes to group difference

estimates. This was true both for imputation based

only on a small number of available covariates

(difference -2.6; 95% CI: -12.0 to 6.9) or for a large

number (difference -0.8; 95% CI: -10.6 to 9.1).

Detailed analysis of the multiply imputed

datasets suggested that variables such as social

class, number of previous episodes of back pain

and presence of pain in the legs at baseline were

significant predictors of reported SF-36 Bodily

pain scores (although adjusting for these factors

made no difference to estimates of group

difference).

Analysis of secondary outcomes

Eighty-two percent of respondents had experienced

at least one episode of back pain within the last 12

months (Table 3).

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.www.acupunctureinmedicine.org.uk/volindex.php 125

Papers

Table 1 Characteristics of respondents and non-respondents, mean (SD) or n/N (%)

Respondents Non-respondentsCharacteristic Acupuncture Usual care Overall Acupuncture Usual care Overall

n=74 n=31 n=105 n=85 n=49 n=134

Age, years 50.3 (10.9) 56.3 (9.5) 52.1 (10.8) 45.5 (10.1) 46.7 (9.4) 45.9 (9.8)

Female 44/74 (60%) 21/31 (68%) 65/105 (62%) 55/85 (65%) 25/49 (51%) 80/134 (60%)

< 14 yrs education 60/74 (81%) 27/31 (87%) 87/105 (83%) 63/81 (78%) 40/48 (83%) 103/129 (80%)

Baseline pain in26/74 (35%) 3/31 (10%) 29/105 (28%) 27/85 (32%) 18/49 (37%) 45/134 (34%)

back not legs

Using pain drugs68/74 (92%) 29/31 (94%) 97/105 (93%) 72/85 (85%) 43/45 (88%) 115/134 (86%)at baseline

No prior episode 11/74 (15%) 6/31 (19%) 17/105 (16%) 14/85 (17%) 7/49 (14%) 21/134 (16%)of back pain

Other majorhealth problem at 19/74 (26%) 11/31 (36%) 30/105 (29%) 25/85 (29%) 14/49 (29%) 39/134 (29%)baseline

Baseline SF-36 n=74 n=31 n=105 n=85 n=49 n=134BP score 28.4 (15.7) 33.0 (18.7) 29.7 (16.7) 32.9 (16.3) 28.8 (17.6) 31.4 (16.8)

3-month SF-36 n=72 n=29 n=101 n=74 n=42 n=116BP score 63.0 (20.8) 62.1 (24.2) 62.7 (21.7) 58.9 (25.0) 50.8 (25.5) 55.9 (23.4)

12-month SF-36 n=72 n=29 n=101 n=75 n=39 n=114BP score 66.1 (25.0) 56.7 (23.3) 63.4 (24.8) 62.1 (26.2) 59.5 (21.6) 61.2 (21.7)

24-month SF-36 n=67 n=26 n=93 n=56 n=33 n=89BP score 69.5 (24.3) 64.1 (23.6) 68.0 (24.1) 65.7 (23.9) 55.9 (23.0) 62.1 (23.9)

Currently working 43/64 (67%) 19/27 (70%) 62/91 (68%) - - -

BP - Bodily Pain dimension of the SF-36

Table 2 Unadjusted mean SF-36 Bodily Pain scores since randomisation

Timepoint Acupuncture Usual Caren mean (SD) n mean (SD)

Baseline (a) 159 30.8 (16.2) 80 30.4 (18.0)

3 months (a) 146 60.9 (23.0) 71 55.4 (25.4)

12 months (a) 147 64.0 (25.6) 68 58.3 (22.2)

24 months (a) 123 68.8 (24.0) 59 59.5 (23.4)

5.5 to 7 years 74 67.7 (24.3) 31 66.7 (21.7)

Differencen points n points

Baseline to 5.5 to 7 years 74 +36.9 31 +36.3

(a) - Source for data is reference13

SD - standard deviation

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 125

Respondents in the acupuncture group were nearly

four times more likely than those in the usual care

group to say that their allocated treatment package

had benefited their current back health and six times

more likely to recommend acupuncture (Table 3).

Both groups scored around 4 on a scale of 1-10

regarding present worry about the back problem

(Table 4).

Several questions pertained to strategies

employed for managing a chronic low back problem.

Taking only the 86 respondents who reported no

back pain in the previous year, there was weak

evidence that those in the acupuncture group (7/13)

were more likely to report using specific back

exercises as prevention than those in the usual care

group (0/6) (difference 53.8%; 95% CI, 7.7 to 76.8).

No other differences were discriminated between

groups on strategies for recurrence prevention and

pain minimisation at this time point.

No difference was detected for the ODI between

groups (difference 0.3; 95% CI -5.5 to 6.2, P=0.9) and

similarly there was no evidence of a difference

between the groups on any other dimension of the

SF-36.

Resource Use

Serious investigations for back pain were rare, with

less than 4% of respondents visiting a hospital

department in the previous 12 months (Table 5).

Continuing treatment was more common with

between 7 to 11% of respondents consulting an

ancillary healthcare practitioner about their back pain.

However while about half of the physiotherapy and

chiropractic accessed was provided on the NHS, nearly

all the acupuncture was paid for out-of-pocket (Table

5). Combining with data previously collected at 12

and 24 months we calculated that 25.8% (8/31) of

the usual care respondents and 23.0% (17/74) of those

in acupuncture group had reported using extra-study

acupuncture for their back pain since randomisation.

We did not collect data for several intervening years

so the actual rate of use is likely to be higher.

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.126 www.acupunctureinmedicine.org.uk/volindex.php

Papers

Table 3 The Back Pain Experience Questionnaire (categorical variables)

Question Acupuncture (%) Usual care (%) Difference (CI) P Oddsn/N n/N ratio

LBP in last year

Yes 61/74 (82.4) 25/31 (80.6)

No 13/74 (17.6) 6/31 (19.4) 1.8% (-14.6 to 18.2%) 0.8

Treatment package has

Helped a great 52/65 (80.0) 15/29 (51.7)deal/little bit

Hasn’t helped at all/ 13/65 (20.0) 14/29 (48.3) 28.3% (7.7 to 48.9%) 0.007 3.7made it worse/don’t know

Recommend acupuncture

Yes 59/65 (90.8) 18/29 (62.1)

No/don’t know 6/65 (9.2) 11/29 (37.9) 28.7% (9.7 to 47.7%) 0.002 6.0

How worried compared with start of trial

Much less /less worried 46/65 (70.8) 15/28 (53.6)

No change/ 19/65 (29.2) 13/28 (46.4) 17.2% (-4.3 to 38.7%) 0.2worried/much more worried

LBP - low back pain; CI - confidence interval of the percentage difference

Table 4 How worried are you about your back problem? (0 is not at all worried and 10 is extremely worried)

Acupuncture Usual care Difference in mean

n mean (SD) n mean SD Difference (SE)(a) 95% CI P

62 3.9 (2.7) 27 4.3 (2.9) -0.4 (0.6) -1.7 to 0.8 0.5

(a) - a negative effect implies a better outcome for the acupuncture groupSD - standard deviation; SE - standard error - CI, confidence interval

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 126

DiscussionStatement of findings

This study is the first to examine outcomes more

than two years after acupuncture treatment for chronic

low back pain. We noted that the usual care group had

now ‘caught up’ with the gains made by the

acupuncture group and at this endpoint we did not

detect any differences between the groups. We found

that those in the acupuncture group were more likely

to say that the treatment package allocated to them

helped their back pain. We didn’t find any major

differences between the groups in the way

respondents managed or prevented their pain or in the

moderate levels of worry experienced about their

condition. Approximately a quarter of respondents in

both groups had used acupuncture to treat their back

pain outside of that provided by the study, 40% of

whom had accessed acupuncture within the previous

12 months.

Strengths and weaknesses

We have found some evidence that a short course of

acupuncture maintains but does not continue to

improve chronic low back pain beyond two years;

however, our results may have been influenced by

low power arising from poor response rate, bias in the

respondent sample or confounding factors.

Our response rate was low and the biggest

potential threat to the validity of our study was that

the respondents were in some way different from

the original sample. Under the assumption that

missing data was missing at random (MAR), we

used multiple imputation to estimate possible effects

due to the missing data. The MAR model cannot

estimate bias caused by non missing at random

(NMAR) effects and, indeed, these effects cannot

be estimated without a model of the missingness

mechanism (which would require a secondary study

for validation). In the absence of a definite

understanding of the missingness mechanism, we

cannot be certain of the extent of respondent bias.

However, we would emphasise that, given the data

available, we have used the most rigorous available

techniques to account for possible bias due to missing

data. Regarding confounding, we acknowledge that

use of acupuncture in a quarter of the control group

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.www.acupunctureinmedicine.org.uk/volindex.php 127

Papers

Table 5 Respondent use of healthcare resources for low back pain in previous 12 months, n (%)

Acupuncture Usual Care Totaln=64 n=29 n=93

NHS hospital visits

Accident and Emergency 0 0 0Hospital outpatients 0 0 0Hospital clinic 2 (3.1%) 1 (3.4%) 3 (3.2%)

Total hospital visits 2 (3.1%) 1 (3.4%) 3 (3.2%)

Physiotherapy

Hospital physiotherapy 3 (4.7%) 0 3 (3.2%)GP physiotherapy 2 (3.1%) 1 (3.4%) 3 (3.2%)

Total NHS physiotherapy 5 (7.8%) 1 (3.4%) 6 (6.5%)Private physiotherapy 2 (3.1%) 3 (10.3%) 5 (5.4%)

Total physiotherapy 7 (10.9%) 4 (13.8%) 11 (11.8%)

Chiropractic

NHS chiropractic 3 (4.7%) 0 3 (3.2%)Private chiropractic 2 (3.1%) 1 (3.4%) 3 (3.2%)

Total chiropractic 5 (7.8%) 1 (3.4%) 6 (6.5%)

Osteopathy

NHS osteopathy 0 0 0Private osteopathy 0 0 0

Total osteopathy 0 0 0

Acupuncture

NHS acupuncture 0 1 (3.7%) 1 (1.1%)Private acupuncture 7 (10.8%) 2 (7.4%) 9 (9.8%)

Total acupuncture 7 (10.8%) 3 (10.3%) 10 (10.8%)

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 127

respondents, along with other potential confounders,

may have obscured any between group differences.

Meaning of our results

The data we have collected adds to the literature on

the natural history of low back pain. At some point

between two to five years after randomisation the

usual care group have improved to the gains seen at

two years in the acupuncture group. We theorise that

exposure to acupuncture speeds recovery from a back

pain episode. Several years after randomisation the

respondents scored 12.2 points lower on the SF-36

Bodily Pain dimension than the age adjusted norm,29

and worry about back health was moderate. This

concurs with other data showing that severity of back

pain fluctuates, and only completely resolves in 10-

31% of cases.3;4;30 A proportion of respondents use

acupuncture as a continuing treatment for their back

pain, the majority of whom pay for it privately. A

concurrent qualitative exploration found that some

of the respondents felt the cost of private acupuncture

treatment was prohibitive, although many believed

acupuncture had been helpful or even pivotal in the

treatment of their pain.15

Unanswered questions and future research

This study raises several interesting questions to be

addressed by future research. We theorise that

acupuncture helped participants to achieve and

maintain a level of pain relief more quickly. However,

we found relatively high utilisation of extra-study

acupuncture in both arms of our trial. Has this extra

acupuncture allowed the usual care group to catch up,

or was the passage of time or other factors responsible?

Our results highlight the question of whether

acupuncture offered as a short course over three months

is an appropriate design for research into a chronic

condition. We recommend that further trials allow

participants to access treatments after an initial course

at ‘point of need’ over a year or more, a design that

may more accurately reflect the actual pattern of use

among back pain patients. Examining the effect of an

intervention over a longer time period will enable long

term treatment strategies to be developed and

compared, an especially important consideration for

chronically fluctuating conditions such as back pain.

Second, we found that social class, area of pain and

chronicity of the condition emerged as predictors of SF-

36 Bodily Pain scores and these factors should be

explored in subsequent research. Third, we found there

was less NHS-provided acupuncture for back pain

than physiotherapy for our respondent sample. We

think the issue of access merits consideration given

the evidence for effectiveness and cost effectiveness for

acupuncture in the short and medium term.13;14;31;32

Finally, more long term studies on acupuncture for

back pain are needed to refute or validate this study’s

findings. To prevent low response rates and plan for an

adequate sample size, planning for long term follow up

should be built into trials at the design stage.

Conflicts of interest

None declared.

Sources of support

The project was supported by a research grant from

BackCare.

AcknowledgementsWe would like to thank all the participants of the

YACBAC trial for contributing to this study.

Reference list1. Dunn KM, Croft PR. Epidemiology and natural history of

low back pain. Eura Medicophys 2004;40(1):9-13.2. Beers MH, Berkow R. The Merck Manual of Diagnosis and

Therapy Section 5. Musculoskeletal And Connective TissueDisorders, Chapter 59. Nonarticular Rheumatism: Low backpain. 2006, 17th Edition.; Available at:http://www.merck.com/mrkshared/mmanual/sections.jsp.(Accessed 13 Sept 2006.).

3. Hestbaek L, Leboeuf-Yde C, Engberg M, Lauritzen T, BruunNH, Manniche C. The course of low back pain in a generalpopulation. Results from a 5-year prospective study.J Manipulative Physiol Ther 2003;26(4):213-9.

4. Von Korff M, Deyo RA, Cherkin D, Barlow W. Back pain inprimary care. Outcomes at 1 year. Spine 1993;18(7):855-862.

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.128 www.acupunctureinmedicine.org.uk/volindex.php

Papers

Summary points

There is some evidence for the effectiveness and costeffectiveness of acupuncture in the short term treatment ofpersistent low back pain

No studies have involved more than two years’ follow up

Up to seven years after an RCT comparing acupuncturewith usual care, we found no differences between thegroups (but the response rate was low)

About a quarter of patients from both groups chose to useacupuncture outside the study, usually paying for it

Trials of acupuncture for back pain should have extendedtreatment and follow up

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 128

5. Clinical Standards Advisory Group. Back Pain. Report of aCSAG committee on back pain. 1994 London: HMSO.

6. Maniadakis N, Gray A. The economic burden of back painin the UK. Pain 2000;84(1):95-103.

7. Thomas K, Coleman P. Use of complementary or alternativemedicine in a general population in Great Britain. Resultsfrom the National Omnibus survey. J Public Health (Oxf)2004;26(2):152-7.

8. MacPherson H, Sinclair-Lian N, Thomas K. Patients seekingcare from acupuncture practitioners in the UK: a nationalsurvey. Complement Ther Med 2006;14(1):20-30.

9. Thomas KJ, Coleman P, Nicholl JP. Trends in access tocomplementary or alternative medicines via primary carein England: 1995-2001 results from a follow up nationalsurvey. Fam Pract 2003;20(5):575-7.

10. Manheimer E, White A, Berman B, Forys K, Ernst E. Meta-analysis: acupuncture for low back pain. Ann Intern Med2005;142(8):651-63.

11. Furlan AD, van Tulder MW, Cherkin DC, Tsukayama H,Lao L, Koes BW, et al. Acupuncture and dry-needling forlow back pain. Cochrane Database Syst Rev 2005, (1)CD001351.

12. van Tulder MW, Furlan AD, Gagnier JJ. Complementaryand alternative therapies for low back pain. Best Pract ResClin Rheumatol 2005;19(4):639-54.

13. Thomas KJ, MacPherson H, Thorpe L, Brazier J, Fitter M,Campbell MJ, et al. Randomised controlled trial of a shortcourse of traditional acupuncture compared with usual carefor persistent non-specific low back pain. BMJ2006;333(7569):623.

14. Ratcliffe J, Thomas KJ, MacPherson H, Brazier J.A randomised controlled trial of acupuncture care forpersistent low back pain: cost effectiveness analysis. BMJ2006;333(7569):626.

15. Esmond LE, Thomas K, Prady SL, MacPherson H. Patientexperiences of longer term effects of acupuncture for chronicpersistent low back pain. In preparation.

16. Ware JE,Jr, Sherbourne CD. The MOS 36-item short-formhealth survey (SF-36). I. Conceptual framework and itemselection. Med Care 1992;30(6):473-83.

17. Brazier JE, Harper R, Jones NM, O’Cathain A, Thomas KJ,Usherwood T, et al. Validating the SF-36 health surveyquestionnaire: new outcome measure for primary care. BMJ1992;305(6846):160-4.

18. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 HealthSurvey. Manual and interpretation guide. Boston: MedicalOutcomes Trust; The Health Institute; 1993.

19. Gilbert FJ, Grant AM, Gillan MG, Vale LD, Campbell MK,Scott NW, et al. Low back pain: influence of early MRimaging or CT on treatment and outcome – multicenterrandomized trial. Radiology 2004;231(2):343-51.

20. Patrick DL, Deyo RA, Atlas SJ, Singer DE,Chapin A,Keller,RB. Assessing health-related quality of life in patients withsciatica. Spine 1995;20(17):1899-908.

21. Walsh TL, Hanscom B, Lurie JD, Weinstein JN. Is acondition-specific instrument for patients with low backpain/leg symptoms really necessary? The responsivenessof the Oswestry Disability Index, MODEMS, and the SF-36.Spine 2003;28(6):607-15.

22. Moore JE, Von Korff M, Cherkin D, Saunders K, Lorig K.A randomized trial of a cognitive-behavioral program forenhancing back pain self care in a primary care setting. Pain2000;88(2):145-53.

23. Fairbank JC, Couper J, Davies JB, O’Brien JP. The Oswestrylow back pain disability questionnaire. Physiotherapy1980;66(8):271-3.

24. Taylor SJ, Taylor AE, Foy MA, Fogg AJ, et al.Responsiveness of common outcome measures for patientswith low back pain. Spine 1999;24(17):1805-12.

25. Roland M, Fairbank J. The Roland-Morris DisabilityQuestionnaire and the Oswestry Disability Questionnaire.Spine 2000;25(24):3115-24.

26. Ostelo RW, de Vet HC. Clinically important outcomes inlow back pain. Best Pract Res Clin Rheumatol2005.19(4):593-607.

27. Little RJA, Rubin DB. Statistical analysis with missing data.2nd ed. New York: John Wiley & Sons, Inc.; 2002.

28. Erdfelder E, Faul F, Buchner A. GPOWER: A general poweranalysis program. Behavior Research Methods, Instruments,& Computers 1996;28(1):1-11.

29. Jenkinson C, Coulter A, Wright L. Short form 36 (SF36)health survey questionnaire: normative data for adults ofworking age. BMJ 1993;306(6890):1437-40.

30. Hestbaek L, Leboeuf-Yde C, Manniche C. Low back pain:what is the long term course? A review of studies of generalpatient populations. Eur Spine J 2003;12(2):149-65.

31. Brinkhaus B, Witt CM, Jena S, Linde K, Streng A,Wagenpfeil S, et al. Acupuncture in patients with chroniclow back pain: a randomized controlled trial. Arch InternMed 2006;166(4):450-7.

32. Witt CM, Jena S, Selim D, Brinkhaus B, Reinhold T, WruckK, et al. Pragmatic randomized trial evaluating the clinicaland economic effectiveness of acupuncture for chronic lowback pain. Am J Epidemiol 2006;164(5):487-96.

ACUPUNCTURE IN MEDICINE 2007;25(4):121-129.www.acupunctureinmedicine.org.uk/volindex.php 129

Papers

AIM December 2007 Text.qxd 21/12/2007 06:38 Page 129