Design overview

Building on our pilot study (ISRCTN32823720) [19], we conducted a parallel-arm randomised controlled trial to determine the effectiveness of acupuncture plus usual care compared to usual care alone for the treatment of IBS based on a published protocol [20]. We received ethics approval from the York NHS Research Ethics Committee (08/H1311/66) in 2008. A cost-effectiveness analysis is reported separately [21].

Settings and participants

We recruited patients from the databases of five general practices. The GP practices identified potential participants aged 18 or older who had consulted their GP and been diagnosed with IBS, and coded accordingly on the practice database. A letter was sent to all potentially eligible participants, inviting them to complete a baseline questionnaire and consent form and return these to the York Trials Unit. We excluded patients who did not speak English, who had a current diagnosis of haemophilia, hepatitis, HIV, or were receiving cancer care, had had major gastrointestinal surgery in the previous six months, were pregnant, had a history of psychosis or substance abuse, or were receiving acupuncture at the time. Patients were recruited if they provided informed consent and scored 100 or more on the IBS Symptom Severity Score (SSS) [22].

Randomisation

We randomised participants equally to receive either a short course of traditional acupuncture plus usual care or usual care alone. The randomisation sequence was computer generated at the University of York by an independent data manager at the York Trials Unit, and fully concealed from the research and administrative staff who then informed participants of their allocation by telephone and subsequently by letter.

Interventions

Nine professional acupuncturists provided the acupuncture, and all were registered with the British Acupuncture Council with at least three years’ post qualification experience, and were working at independent clinics. Their practice style was based on principles of Traditional Chinese Medicine or “TCM”. Acupuncturists provided up to 10 sessions of acupuncture, adapted from a previously tested protocol [19], which allowed explanations and life-style advice based on acupuncture theory and clinical judgement [16]. While actual selection of points was individualised for patients and allowed to change over time, there was standardisation of the function of the acupuncture based on the theoretical frameworks used, an approach proposed for maintaining interventional integrity when delivering complex interventions [23].

All patients remained under the care of their general practitioner, and received usual care according to need. We documented usual care, for both NHS and non-NHS treatments, in both groups at three, six, nine and 12 months.

Outcomes and follow-up

Our primary outcome measure was the IBS Symptom Severity Score (IBS SSS). This is scored from 0 to 500 (<75 = no IBS, 75–175 = “mild” case, 175–300 = “moderate” and 300+ = “severe”), and it has been validated for use in IBS patients [22]. Our primary end-point was at three months, with further follow-ups at six, nine and 12 months. We recorded participants’ baseline belief in acupuncture, expectations that acupuncture might help their IBS and treatment preferences.

Our secondary outcome measures were the IBS Non-Colonic Symptom Score (which includes lethargy & tiredness, “wind”, backache, and other symptoms [24], and SF-12 to evaluate patients’ health related quality of life [25]. EQ-5D was the primary outcome for the economic study [26]. Outcomes at follow-up points were sought by postal questionnaire, and where that failed to elicit a response, the main outcome measure was sought by telephone.

Statistical analysis

In our pilot study [19], we calculated that for a 50 point difference on the IBS SSS, and an adjusted standard deviation to be 105 points, we needed to recruit a sample size of 94 patients per arm to have 90% power and a significance level of 0.05. To allow for loss to follow up of a similar proportion as observed in the pilot at three months (13%), our trial required at least 220 patients. In the full-scale trial we used analysis of covariance on an intention-to-treat basis to evaluate changes between groups for the primary outcome of the IBS SSS at three months, while adjusting for baseline scores and general practice. The statistical analysis was conducted using SAS version 9.2 and STATA version 10 using a 2-sided 5% significance level. We took a 50 point improvement [22] or more on the IBS SSS between baseline and three months as a “success” for our number-needed-to-treat analysis. We used similar analyses to explore between-group differences on the IBS SSS at other time points, and on secondary outcome measures. We used linear regression to explore whether expectation, belief and preference were treatment effect modifiers.

Recruitment and follow-up of participants

The trial recruited patients from five general practices (total list size of 53,666) between November 2008 and June 2009. GP practices identified and invited 1,651 potentially eligible patients to participate. Of the 243 patients who responded with baseline documentation to the York Trials Unit, 233 were recruited, see Flow Chart in Figure 1. The average number recruited per general practice was 47 (range 24 to 75).

Baseline characteristics

Acupuncture provision within the acupuncture arm of the trial

Usual care provision across both arms of the trial

We mapped the provision of usual care in both groups, which was provided or purchased according to need, and found differences in utilisation were minimal (see Additional file 1). For example in the first three months 59/83 (69%) of patients in the acupuncture group consulted at the their GP practice, compared to 72% of patients in the usual care group, with patient consulting on average 2.08 times and 2.25 times respectively, with 68% and 69% of these visits being related to IBS respectively. Over the first three months, 65% of the acupuncture group received prescription medication compared to 61% of the usual care group, with antispasmodics drugs, antidiarrhoeal drugs and laxatives making up a relatively small proportion of this utilisation (Additional file 1). Non-prescription medication was used by 58% in the acupuncture group compared to 60% in the usual care group.

Patient outcomes

Table 3 shows the results from the linear regression model providing IBS Symptom Severity Scores at baseline, 3, 6, 9 and 12 months, as illustrated in Figure 2. The between-group comparison showed a statistically significant reduction in symptoms in those allocated to acupuncture at three months: –27.43 (95% CI: –48.66 to −6.21, p = 0.012). The effect size could be described as small to moderate (Cohen’s d = 0.36). Treatment was assumed “successful” (≥50 points reduction on IBS SSS) in 57/116 (49%) of patients in the acupuncture group and 36/117 (31%) in those receiving usual care alone, a difference of 18% (95% CI: 6% to 31%). The number needed to treat was 6 (95% CI: 3 to 17), which means that for roughly every 6 participants treated with acupuncture, you will see one participant with an improvement equal to or greater than 50 in their baseline IBS SSS score.

Variable	Baseline Mean (SD)	Month 3 (95% CI)	Month 6 (95% CI)	Month 9 (95% CI)	Month 12 (95% CI)
Primary analysis: IBS Symptom Severity Score
Acupuncture group	280.00 (81.34)	213.78 (199.31 to 228.25)	201.60 (186.08 to 217.12)	206.40 (190.39 to 222.41)	209.79 (194.56 to 225.03)
Usual care group	277.17 (71.50)	241.21 (225.69 to 256.73)	224.19 (207.91 to 240.48)	233.40 (217.21 to 249.59)	231.12 (214.81 to 247.44)
Between group differences in means (acupuncture-usual care)	-	−27.43 (−48.66 to −6.21)	−22.59 (−45.11 to −0.08)	−27.00 (−49.77 to −4.23)	−21.33 (−43.66 to 1.00)

At each time point, predicted means and their 95% confidence intervals from the fitted model are presented except at baseline where the raw means and their standard deviation are presented.

The difference on the IBS SSS scores between groups subsequent to 3 months, adjusted for baseline, showed a statistically significant reduction at 6 months [−22.59] and 9 months [−27.0], followed by a near significant reduction at 12 months [−21.33], see Table 3 for confidence intervals. We found that patient expectation and belief were not treatment effect modifiers. Nor was preference, whether on the basis of comparing groups who preferred acupuncture vs. those who didn’t mind, or when comparing those who got their preference vs. those who did not. We found a statistically significant reduction favouring the acupuncture group for the IBS Non-Colonic Score only at three months, [−15.24 (95% CI: –29.91 to −0.57)], but no significant differences in the physical and mental component scores on the SF-12 at any time point (see Additional file 2).

Adverse events

No serious adverse events associated with acupuncture were reported by acupuncturists in the logbooks or by patients in follow-up questionnaires. Non-serious adverse events associated with acupuncture were reported by eleven patients: eight experienced pain from the needles (three of the eight also reported nerve pain, bruising or fatigue), one had bruising from the needles, one had a slight burn from a moxibustion stick, and one had itchy and swollen skin where one of the needles had been inserted.

Summary of main findings

For patients with chronic irritable bowel syndrome, our study showed that acupuncture provided as an adjunct in primary care is associated with a small yet statistically significant clinical reduction in symptoms at three months. This was reflected in 49% of patients in the acupuncture group and 31% in the usual care group improving by more than 50 points in their IBS SSS score, with six being the number-needed-to-treat. This effect was largely sustained through to 12 months. We also found acupuncture for this group of patients had a high level of acceptability, as shown by the willingness by patients to attend for acupuncture once they had commenced treatment, completing on average 9 out of the 10 weekly acupuncture sessions that were available.

Strengths and limitations

This study is the first pragmatic randomised controlled trial of acupuncture for IBS in primary care. The design ensured that we controlled for temporal effects associated with spontaneous recovery and regression-to-mean effects associated with having a minimum cut-off on our IBS SSS scale. The trial had an adequate sample size, and was conducted rigorously at all stages, with unambiguously concealed allocation to prevent subversion of the randomisation. Of the 1,651 potentially eligible patients identified by GP practices that were invited to participate, 243 patients responded, a rate that is typical of trials in primary care that use a database method for recruitment. We used validated outcomes measures, with response rates at 3, 6, 9 and 12 months that were reasonably high, at over 80 per cent in the control group and over 90 per cent in the intervention group. We have documented that there was reasonable equivalence in usual care received in both groups at three months, which leads us to cautiously ascribe any differences between groups to the treatment provided by acupuncturists. We explored the impact of variables that are potentially associated with placebo effects, namely patients’ beliefs, expectations and treatment preferences [29], but they were not found to be treatment effect modifiers.

There is likely to be a contribution to the overall benefit we observed due to the additional attention received in the acupuncture group resulting from the contact time with practitioners in the acupuncture arm. We did not control for this bias, which would have required a trial arm with a practitioner-led intervention in order to control for time and attention. The patients in our study were not blind to their group allocation, as would have been the case in a sham-controlled trial. A comparator trial arm with a sham control would have had the limitation that the verum acupuncture would need to be constrained to minimise variability in patient experience between the verum and sham acupuncture arms. There is a trade-off between seeking to know more about the contribution each of the components might (or might not) contribute to any putative benefit and seeking to know more about the impact of acupuncture as a package of care that more closely reflects routine practice. A third trial arm was outside of the scope of our funding support.

The study population was made up of those who have consulted with predominantly “moderate” IBS symptoms in primary care (average IBS SSS score of 279), have experienced these symptoms for an average of 13 years and have not experienced sufficient symptom reduction from their usual care so that they are continuing to seek care. Therefore generalising the results to patients who are consulting their GP for the first time, and who therefore have not experienced usual GP care and its impact on symptoms, must be done with caution. The trial was not powered to determine whether any of the sub-groupings of patients with IBS symptoms (for example based on severity) would have fared better or worse.

Comparison with existing studies

Our study had the longest duration in terms of collecting outcome data through to 12 months after randomisation. By contrast, most other studies have been conducted within hospital based settings, with more narrowly defined populations, smaller sample sizes and shorter term follow-ups [4]. The level of acceptability of acupuncture treatments among patients was high (only three patients out of 116 did not start treatment and 88 (76%) completed their full course of treatment, with an overall average of nine sessions attended out of 10) compared to a trial of cognitive behavioural therapy for IBS in which fewer than half of patients were considered by the therapist to have completed therapy, with 41% either declining therapy or dropping out [30].

Components of acupuncture are drawn from three categories: needling related components; non-needling components that are acupuncture-specific, such as explanations and lifestyle advice; and non-specific components that are generic to the therapeutic encounter, such as the patient-practitioner relationship [31]. Our trial is unique in that it was designed to include the second of these components, as well as the other two. However it was not designed to determine the extent to which each of these components of acupuncture contributed to the overall outcome. To establish a proportional contribution of a component of treatment would require a trial arm with a sham control, for example as has been done recently to explore the impact of the patient-practitioner relationship [10]. In their trial, sham acupuncture needling was used to control for the needling in verum acupuncture arm, which then allowed the impact of a variation in the intensity of patient-practitioner relationship to be measured. By contrast, our focus on acupuncture as a complex intervention with many interacting components meant that a feasible and valid sham control was problematic [16, 17].

With regard to placebo effects among patients with IBS, there is some limited evidence from other studies that these effects appear to tail off after three months, according to a study that reviewed 27 randomised controlled trials of treatments for IBS, with individual trials showing no placebo response at 12 months [32]. The results from our trial, in contrast, provide some evidence of a sustained effect at 12 months.

Implications for future research and clinical practice

Our trial focussed on those patients with long term symptoms at recruitment (average 13 years duration) for whom usual GP care was of limited benefit; however, exploring the potential impact of referral to acupuncture soon after an initial diagnosis of IBS would be useful. An appropriately powered study could determine whether any of the sub-groupings of patients with IBS symptoms (for example based on mild, moderate or severe categories) fared better or worse, and also characterize responders and non-responders to acupuncture. A clearer specification of optimal acupuncture treatment for this group of patients would be helpful. In particular it would be useful to determine if more than 10 acupuncture sessions would improve outcomes, as from the graph in Figure 2 the steep improvement of the acupuncture group appears to level off quite abruptly when treatment ceases. To separate out the impact of different components of treatment, a sham controlled trial design similar to that used by Kaptchuk and colleagues [10] could be used to determine whether or not the provision of explanations and life-style advice might (or might not) add to any overall effect, assuming that effects of different treatment components are additive rather than synergistic.

There is a need for a number of effective options for treatment in primary care, given the frustration experienced by some patients with existing treatments [2]. Our results showed that, on average, acupuncture is associated with a small to moderate reduction in symptoms in a population that has had IBS symptoms for many years. If a patient responds at three months, then this benefit is likely on average to be sustained at 12 months. The IBS Symptom Severity Score is now widely used in IBS research and has been translated into many languages. When used in assessing response to treatment, a reduction of 50 points is regarded as clinically significant and, therefore, a reduction of 70 points after 12 months in a group of primary care patients in whom symptoms have been refractory, suggests that this is a treatment option worth considering in such individuals, especially as the beneficial effects seem to be sustained. These data on acupuncture are highly relevant to the primary care context in the UK and have policy implications.