This is a review of randomised trials comparing acupuncture Vs no acupuncture.

We sought trials that compared acupuncture to usual back pain care. We didn’t restrict age, sex or types of acupuncture. We didn’t consider trials of cost effectiveness, or comparing acupuncture with other complimentary therapies.

We conducted a thorough search through major databases, i.e. MEDLINE, EMBASE and Allied and Complimentary medicine and managed to retrieve 274 articles. Our search terms were acupuncture, lumbar and lumbosacral region, pain, backache and low back pain. Our limitations were articles in English language and between the years 2003 and 2005. Review of abstracts identified 8 randomised controlled trials. Last Cochrane review was in 2004 on this topic.₁₀ Out of 8, 3 articles were not focused on the topic and were excluded. ₁₁,₁₂,₁₃,₁₄

Figure 2

Table 2 Comparison of treatment interventions

Two studies were funded by insurance and one by NHS (government) and one was funded by Arthritis Association (New York Chapter)

All 4 studies were multi centred trials conducted in United States of America, United Kingdom and Germany. Investigators selected different age groups of adult population from 18 to 75 years of age with history of chronic low back pain more than 4 weeks to 8 years. They excluded patients with spinal tumour, infection, and fracture or with any neurological deficit (table 1) or spinal cord compression or deformity of the spine.

All 4 studies used computer-generated randomisation to allocate patients into treatment groups. Baseline characteristics were well balanced. Haake et al used 1:1:1 but rest of the investigators preferred to 2:1 randomisation to balance the skills of acupuncturists. All of these studies were based on questionnaires and telephone interviews. Thomas et al couldn’t blind their study because of “pragmatic reasons” but rest of three studies were blinded. Haake et al and Meng et al also assessed their blinding in their questionnaires. It is not very clear whether all patients received their allocated sessions or not. In general, 10/337 in Haake’s₁₇, 1/160 in Thomas’₁₈, 27/147 in Brinkhaus’¹⁹, 3/28 in Meng’s²⁰ study didn’t get the active acupuncture treatment. Whether they were included in the final intention to treat analysis is not entirely clear except in Haake’s study where unblended patients were considered non-responders.

All 4 studies compared acupuncture Vs no acupuncture. Haake et al and Brinkinhaus et al compared 3 groups namely acupuncture, sham (minimal acupuncture) and no acupuncture (conventional therapy). However, Meng et al and Thomas et al didn’t consider sham acupuncture since it is superficial needling at non-acupuncture points.

All 4 studies measured multiple outcomes using validated tools for back specific disability, pain, general function, and psychological function (Table 1).

Haake used Von Korff Chronic Pain Grade Scale (VKCPGS) Questionnaire where 33% improvement or 3-point improvement considered statistically significant or 12% improvement on back-specific Hanover Functional Ability (HFA) Questionnaire. This study proved 3.4% (CI –3.7-10.3%; P value-0.39) improvement in Verum vs. sham acupuncture. However, Verum vs. conventional therapy shown considerable improvement of 20.2%(CI13.4%-26.7%; P value-0.001) and sham vs. conventional therapy were 16.8%(CI 10.1%-23.4%; P value 0.001).

Meng et al used Roland Disability questionnaire as main outcome measure where 4-point reduction is significant. They managed to get 4.1-point reduction at week 6 Vs 0.7 point with conventional therapy. He also used global transition score.

Figure 3

Table 1 Flow chart of screening, randomisation, Lost follow up and final analysis.

Brinkhaus et al used modified version of pain questionnaire published by German society for study of pain but primary outcome measure was Visual Analogue Scale (VAS). 40 points change considered significant in VAS. They found 28.7 30.2 mm improvement in acupuncture group, 23.631mm in minimal (sham) acupuncture and 6.922 mm in waiting list patients who were on conventional therapy.

Thomas et al used SF-36 (main outcome measure), oswestry disability index (ODI), McGill present pain index, safety and patient satisfaction. At 12 months, they found 30.8 points reduction in acupuncture and 30.4 points in patients on conventional therapy. This interventional effect was statistically more significant at 24 months at 8 points between acupuncture and no acupuncture (>5 points is significant in SF-36) as compared to 5.6 points at 12 months. Therefore, shown strong evidence of small effect at 24 months but week evidence of some effect of acupuncture at 12 months.

Haake et al study is one of the biggest and vigorous trials carefully conducted on acupuncture in chronic low back pain. They managed to prove considerable difference between acupuncture vs. guideline based conventional therapy. They managed to segregate physiologic effect from psychological effect by comparing Verum with sham acupuncture which further accentuate the credibility of the trial and understanding of the way acupuncture works. Positive points would be high power, low drop out rates and blinding checks. Negative points can be overtly poor results with conventional therapy.

Meng et al study is the smallest of all 4. Few positives are patients being sole assessors of their symptoms; similarly cross over of patients after the completion of the study did prove the symptom improvements after the treatment. It also reduces the confounding of acupuncture and same pain relief as of acupuncture group further confirms the findings of the acupuncture group. Negative points were low power, needle improvement was mixed with electric stimulation, patients with buttock and leg pain (? Spinal/ vascular) weren’t excluded. On the other hand, 3 patients were excluded because they couldn’t fill in the forms properly and finally, high loss of follow up.

Brinkhaus et al again presented with similar results. They managed to prove statistically significant difference between acupuncture and no acupuncture but no difference between two types of acupuncture. Positive points of this study are well-balanced baseline characteristics and again cross over to prove the results further. It was big multicentre trial but was conducted on outcome measure of limited sensitivity and specificity. They later also took sham acupuncture as placebo but later proven to be wrong and ethically patients were informed of kind of acupuncture they will get after being randomised, though this did help in blinding.

Thomas et al study produced few surprising results that acupuncture produces more long-term results than short term. Technically, well organises study with well-predicted drop out rates and compensating power analysis. 2:1 randomisation allowed to accommodate the difference of acupuncturist’s skills but this did tripped the balance towards acupuncture. Few negatives include open (unblended) trial, “individualizing” the patients (extra 5 acupuncture sessions was provided if pain is not better), all “off work” patients were offered acupuncture and statistically low power but then re recruiting of patients to improve the power.

In conclusion, none of these studies could confirm dramatic improvements in chronic low back pain after acupuncture therapy in short or long term despite of big power but few of them managed to reach statistically significant confidence intervals.