In October 2000 the German Federal Committee of Physicians and Health Insurers recommended that special Model Projects on Acupuncture (“Modellvorhaben Akupunktur”) be developed in order to determine the evidence-based role of acupuncture in the treatment of certain illnesses. This paper presents a summary of the main randomised controlled trials performed as part of these projects, and the associated economic analyses.
Overall the results show that acupuncture is effective in practice for a range of chronic conditions, and it seems likely to have acceptable cost utility (at least at a rate of €35 per session). Sham acupuncture, in the form of minimal off-point needling in a therapeutic context, also appears to be effective, being no different to prophylactic medication in migraine, and superior to guideline-based standard care in chronic low back pain.
In patients recruited to acupuncture trials, the response to treatment does not differ between those that agree to be randomised and those that do not. This suggests that the results of the pragmatic Acupuncture in Routine Care studies are applicable to patients from the general population who express a preference for acupuncture.
In conclusion, acupuncture appears to be effective in a range of chronic conditions and it seems to have acceptable cost-effectiveness in Western health economic terms. These programmes of research do not confirm the hypothesis that needling at specific points is essential to achieve satisfactory clinical effects of acupuncture. Sham acupuncture, in the form of minimal off-point needling in a therapeutic context, is unlikely to be an inactive placebo.
In April 2006, the German health authorities decided that acupuncture would be included into routine reimbursement by social health insurance funds for chronic low back pain and chronic osteoarthritis of the knee.
Statistics from Altmetric.com
Three large research programmes investigating the efficacy, effectiveness, cost-effectiveness and safety of acupuncture treatment for certain chronic conditions (“Modellvorhaben Akupunktur”) have been conducted in Germany since October 2000.1 These programmes were initiated after the German Federal Committee of Physicians and Health Insurers determined, in October 2000, that the scientific evidence supporting the use of acupuncture was not sufficient to justify routine reimbursement within the German healthcare system. Formerly, some of the cost of acupuncture treatment was covered by the German statutory health insurance funds, provided that the acupuncture was performed by physicians with at least 140 h of acupuncture training.
Following the decision of the German Federal Committee of Physicians and Health Insurers in October 2000, reimbursement for acupuncture treatment was only possible for patients suffering from certain chronic conditions (knee osteoarthritis, low back pain, migraine, tension-type headache), and only if the physician performing the acupuncture participated in one of the three research programmes. The results discussed in this paper are from the controlled trials that made up the core of these research programmes: the Acupuncture Randomised Trials (ART),2–5 the Acupuncture in Routine Care (ARC) studies,6–14 one comparative trial (COMP),15 and the German Acupuncture trials (GERAC).16–19
These Modellvorhaben (trial phases) were funded by a number of the German statutory health insurance funds, and they were organised by groups of researchers and physicians based at three large German universities: the Charité University, Berlin (ART and ARC); the Technical University, Munich (ART and COMP); and the University of Bochum (GERAC).
These were four randomised controlled trials (RCTs) with roughly 300 subjects in each. They were performed principally as efficacy trials in four conditions: migraine3; tension-type headache4; chronic low back pain2; and osteoarthritis of the knee.5 Each of the trials followed the same design: three parallel arms with a 2:1:1 distribution of subjects, so that there were approximately 150 subjects in the real (verum) acupuncture arm, and 75 in the others — the minimal acupuncture and waiting list arms. The acupuncture involved deep needling to classical acupuncture points with manipulation of the needles to produce de qi — a characteristic needling sensation. Twelve treatments were given over 8 weeks. Minimal acupuncture involved superficial needling to standardised sites that were not near to any recognised acupuncture points. The waiting list group received acupuncture 2 or 3 months after randomisation, that is, after the data were collected for the primary outcome.
The primary outcomes were short term, just after the interventions at around 8 weeks, although outcomes were also assessed at 26 and 52 weeks from baseline.
The primary outcome for ART migraine was the difference in number of days with headache of moderate or severe intensity between the 4 weeks before randomisation (baseline phase) and weeks 9–12 after randomisation. Responders were defined (post hoc) as those with a 50% reduction or greater in days with moderate or severe pain (headache). The primary outcome and responder rates for ART tension-type headache were the same, with an additional comment that patients with missing data were automatically counted as non-responders.
The primary outcome in ART low back pain was the change in low back pain intensity from baseline to the end of week 8 after randomisation, as measured by a visual analogue scale (range, 0–100 mm), and responders were defined (post hoc) by at least 50% reduction in pain intensity. Finally, the primary outcome measure in ART knee osteoarthritis was the change in the Western Ontario and McMasters Universities Osteoarthritis Index (WOMAC) between baseline and week 8 after randomisation, and responders were defined (post hoc) by a decrease of at least 50% in their WOMAC index score.
The ART programme trials were performed across between 18 and 30 outpatient centres across Germany: ART migraine 18; ART tension-type headache 28; ART low back pain 30; and ART knee osteoarthritis 28.
This was a single comparative trial of acupuncture and metoprolol (100–200 mg) in migraine prophylaxis.15 It was designed as an equivalence trial, with a recruitment target of 480. The acupuncture treatment consisted of at least eight to a maximum of 15 sessions of 20 to 30 minutes’ duration, administered over a period of 12 weeks. At least six needles were used per session, with manual stimulation at least once to achieve de qi. The metoprolol intervention was based on the recommendations of the German Migraine and Headache Society and consisted of metoprolol 100–200 mg daily for 12 weeks.
The primary outcome measure was the difference in number of days with migraine between the 4 weeks before randomisation (baseline) and weeks 9–12 after randomisation as reported by the patient in a standardised headache diary. Responders were defined as those with at least a 50% reduction in migraine attacks.
There were four GERAC trials with up to 1000 subjects in each. They were designed as comparative trials with three equal parallel arms: acupuncture versus sham acupuncture versus standard care (note that the terminology “sham” is used rather than “minimal”). They were performed in migraine,18 tension-type headache,16 chronic low back pain,17 and knee osteoarthritis.19 Rather like the ART trials, the real acupuncture involved deep needling to classical acupuncture points with manipulation of the needles to produce de qi. Ten treatments were given over 6 weeks, with the option to extend treatment by a further five sessions for partial response. Sham acupuncture involved superficial needling to standardised sites that were not near to any recognised acupuncture points. The standard care arms used best conventional care based on guidelines where available: GERAC migraine — beta blockers first choice, flunarizine second, valproic acid third; GERAC tension-type headache — the intention was to use amitriptyline; GERAC chronic low back pain — multimodal treatment programme including physiotherapy, exercise and non-steroidal anti-inflammatory drugs (NSAIDs); GERAC knee osteoarthritis — physiotherapy, physician visits, NSAIDs (in this trial all groups had six sessions of physiotherapy, and acupuncture groups were allowed limited NSAIDs as rescue medication).
The primary outcomes were measured around 6 months from baseline, although secondary outcomes were measured at 6 weeks and 3 months as well.
The primary outcome in GERAC migraine was the difference in migraine days between 4 weeks before randomisation and weeks 23–26 after randomisation, and response was defined as a reduction in the number of migraine days by 50% or more. GERAC tension-type headache was somewhat different in that the response (defined as >50% reduction in number of headache days per 4 weeks from baseline to 6 months) was the primary outcome, and all minor variations from protocol resulted in patients being classified as non-responders.
In GERAC low back pain the primary outcome was response after 6 months, defined as 33% improvement or better on three pain-related items on the Von Korff Chronic Pain Grade Scale questionnaire (CPGS) or 12% improvement or better on the back-specific Hanover Functional Ability Questionnaire (HFAQ). Patients who were unblinded or who used (disallowed) co-interventions during follow-up were classified as non-responders regardless of symptom improvement.
In GERAC knee osteoarthritis the effect on pain and function was measured with the WOMAC score (total score and the subscales were standardised to 0–10). “Success” rates were calculated according to a change of at least 36% from baseline WOMAC scores at 13 and 26 weeks after the start of treatment. Patients with missing data were considered to have had treatment failure.
The GERAC trials were performed across between 122 and 340 practices across Germany: GERAC migraine 149; GERAC tension-type headache 122; GERAC chronic low back pain 340; GERAC knee osteoarthritis 315.
The ARC studies were a series of large to very large pragmatic RCTs, with associated non-randomised cohorts. They used a standard design and included detailed economic analysis from a societal perspective. Subjects insured by one of the participating social health insurance funds were recruited by general practitioners across Germany for acupuncture treatment of either: osteoarthritis of the hip or knee712; chronic neck pain69; chronic low back pain8; chronic headache1113; dysmenorrhoea14; allergic rhinitis10; or asthma (awaiting publication). If they agreed to be randomised, they either received 15 sessions of manual acupuncture over 3 months, or they waited 3 months for acupuncture treatment. If they expressed a strong preference for acupuncture and declined to be randomised, they received acupuncture treatment immediately. There was no standardisation of treatment, but only manual acupuncture was allowed.
Outcomes were measured at 3 and 6 months. After 3 months the group randomised to usual care alone were given acupuncture treatment. The primary outcomes were all set at 3 months. ARC chronic headache used the reduction in days with headache per month. ARC low back pain measured back function assessed by the HFAQ. ARC osteoarthritis used the change in WOMAC score, and ARC chronic neck pain used a validated neck pain and disability scale (NPAD). In ARC dysmenorrhoea the main outcome was the average pain intensity during the last menstruation before assessment measured on a numeric rating scale. ARC allergic rhinitis used the Rhinitis Quality of Life Questionnaire (RQLQ).
As part of the ARC programme of studies, additional measurements were performed to assess quality of life (QoL), costs and the cost-effectiveness relationship of routine care plus acupuncture compared with routine care alone. QoL was assessed with the Short Form (SF)-36 questionnaire, using the subscales and the components scales. The SF-36 also served as the basic benefit estimator for the cost-effectiveness analyses. At baseline and at 3 months the patients completed questionnaires which assessed the QoL over the previous 7 days. The costs considered were measured in societal perspective and included the direct healthcare-related costs of acupuncture (cost of each acupuncture session was €35), physician visits and hospital stays, and any drugs prescribed. In addition to health insurance costs, the indirect costs caused by lost workdays were also taken into account. These were estimated to be approximately €78 per lost workday. Additional analyses were performed to estimate cost utility in the case of higher costs and better medical outcome. QoL measures using SF-36 were converted to quality-adjusted life-years (QALYs), and the excess cost in the acupuncture group in each study was divided by the increment in QALYs gained in the acupuncture group compared with the usual care group. This gave an incremental cost-effectiveness ratio (ICER) expressed as a cost per additional QALY.
An overview of the main results of the ART trials is shown in fig 1, expressed as responder rates for comparison across the different conditions. In all four trials there were significant short-term differences between acupuncture and waiting list, but there was a significant difference between acupuncture and minimal acupuncture only in ART knee osteoarthritis. There were significant short-term differences between minimal acupuncture and waiting list in all four trials. Treatment effects were maintained in the acupuncture and minimal acupuncture groups at long-term follow-up (21–24 weeks in ART migraine and tension-type headache; 52 weeks in ART low back pain and knee osteoarthritis).
Recruitment for this trial proved very difficult and the trial was ended prematurely after 114 patients had been randomised. Two of 59 patients randomised to acupuncture withdrew prematurely from the study compared with 18 of 55 randomised to metoprolol. The number of migraine days decreased by 2.5 days in the acupuncture group compared with 2.2 days in the metoprolol group (p = 0.721). The proportion of responders (reduction of migraine attacks by ⩾50%) was 61% for acupuncture and 49% for metoprolol. Responder rates are shown on the left of fig 2. Both physicians and patients reported fewer adverse effects in the acupuncture group.
An overview of the main results of the GERAC trials is shown in fig 2. These are expressed as responder rates. Responder rates were defined as: ⩾50% reduction in migraine days (GERAC migraine); >50% reduction in headache days (GERAC tension-type headache); ⩾33% improvement on CPGS or ⩾12% improvement on the HFAQ (GERAC low back pain); and ⩾36% improvement in WOMAC (GERAC knee osteoarthritis). Note that in GERAC tension-type headache, half of the subjects with greater than 50% reduction in headache days were classified as non-responders due to medication changes, use of co-interventions, protocol violations or unblinding.
The primary outcome in GERAC migraine showed a mean reduction of 2.3 days in the acupuncture group, 1.5 days in the minimal acupuncture group, and 2.1 days in the standard therapy group. These differences were statistically significant compared with baseline (p<0.0001), but not across the treatment groups (p = 0.09).
The primary outcome in GERAC tension-type headache was the response — in the intention-to-treat analysis (all 409 patients), 33% of patients in the acupuncture group and 27% of patients in the minimal acupuncture group (p = 0.18) were classed as responders. Acupuncture was superior to minimal acupuncture for most secondary outcomes, including headache days (1.8 fewer; p = 0.004) and the International Headache Society response criterion (>50% reduction in headache days: 66% vs 55%, risk difference 12%; p = 0.024).
In GERAC low back pain the response rates at 6 months were 47.6% (acupuncture), 44.2% (minimal acupuncture), and 27.4% (conventional therapy). The differences among groups were as follows: acupuncture versus minimal acupuncture, 3.4% (p = 0.39); acupuncture versus conventional therapy, 20.2% (p<0.001); and minimal acupuncture versus conventional therapy, 16.8% (p<0.001).
In GERAC knee osteoarthritis the success rates were 53.1% for acupuncture, 51.0% for minimal acupuncture, and 29.1% for conservative therapy. Acupuncture groups had higher success rates than the conservative therapy group (relative risk (RR) for acupuncture compared with conservative therapy, 1.75 (p<0.001); RR for sham acupuncture compared with conservative therapy, 1.73 (p<0.001)). There was no difference between acupuncture and minimal acupuncture (RR, 1.01 (p = 0.48)).
An overview of the primary outcomes of the ARC trials is shown in fig 3. These results are expressed as percentage changes in the primary outcome measures. The results are shown at 3 months in all groups apart from the “control after acupuncture” (green bar, third from the left in each set of results). The “control after acupuncture” represents the results after acupuncture was provided for the group initially randomised to “usual care control”. Comparisons cannot be made between the percentage changes across different studies, but the patterns can be compared. The numbers of participants included in each trial are noted below each set, and expressed as two numbers: the number included in the randomised part of each trial followed by “r”; and the total number included in the whole trial, including the large non-randomised cohort. In the randomised part of each trial, there were clinically relevant differences at 3 months between the groups receiving acupuncture plus usual care and those receiving usual care alone. These differences were very highly significant (p<0.001). The non-randomised groups tended to have higher symptom severity at baseline, but there were no significant differences in terms of the response to acupuncture when compared with the randomised acupuncture group.
The associated cost-effectiveness analysis results are shown in fig 4. These are ICERs at 3 months from baseline, and are expressed as costs per additional QALY in the group randomised to acupuncture.
The Modellvorhaben Akupunktur include the largest clinical studies on acupuncture ever performed, and both the methods and the results have caused considerable debate. But they are a tremendous achievement, and the researchers involved should be congratulated for their efforts.
The Modellvorhaben Akupunktur are three large research programmes that have attempted to investigate the efficacy, effectiveness, cost-effectiveness and safety of acupuncture treatment for certain chronic conditions including headache, back pain, knee pain and neck pain
The results demonstrate that acupuncture is effective in practice
Acupuncture (at a rate of €35 per session) in addition to usual care seems likely to have acceptable cost utility when compared with usual care alone in these conditions
Sham acupuncture, in the form of minimal off-point needling in a therapeutic context, appears to be no different to prophylactic medication in migraine, and superior to guideline-based standard care in chronic low back pain, hence it is unlikely to be an inactive placebo
In patients recruited to acupuncture trials, the response to treatment does not differ between those that agree to be randomised and those that do not
Length of training in acupuncture does not seem to influence the results of treatment
Overall the results show that acupuncture is effective in practice for a range of chronic conditions, and it seems likely to have acceptable cost utility (at least at a rate of €35 per session). Sham acupuncture, in the form of minimal off-point needling in a therapeutic context, also appears to be rather effective, being no different to prophylactic medication in migraine (GERAC migraine), and being superior to guideline-based standard care in chronic low back pain (GERAC low back pain). This challenges the assumption that minimal, off-point needling is a placebo.
In view of the results of acupuncture over sham, perhaps it is not a surprise to learn that the length of training of the acupuncturists did not have any significant influence on the results of treatment. However, it is interesting to note that the effect in the sham group in GERAC knee osteoarthritis appears to be much larger than that in ART knee osteoarthritis, even allowing for the difference in the percentage used to define response. Perhaps it is more important to be properly trained in performing sham acupuncture for RCTs than in therapeutic acupuncture in practice. It should also be noted that the sham acupuncture group in GERAC knee osteoarthritis also received six sessions of physiotherapy, which may partly explain the larger observed effect in this trial.
In patients recruited to acupuncture trials, the response to treatment does not differ between those that agree to be randomised and those that do not. This suggests that the results of the ARC studies are applicable to the general population, provided that they are willing to try acupuncture, that is, express a preference. It should be noted, however, that recruitment to the Modellvorhaben Akupunktur was significantly aided by economic pressures — in the whole of Germany reimbursement for the cost of acupuncture treatment was only available to patients who entered these trials.
In conclusion, manual acupuncture performed by German physicians with at least 140 h of training is effective in a range of chronic conditions, and the cost utility appears to be acceptable in Western health economic terms. These programmes of research do not confirm the hypothesis that classical needling techniques at specific points are essential to achieve optimal effects of acupuncture. Sham acupuncture, in the form of minimal off-point needling in a therapeutic context, is unlikely to be an inactive placebo.
Thanks to Claudia Witt and Benno Brinkhaus for providing data where necessary to complete the presentation of results.
Competing interests: MC is medical director of the BMAS. This role involves running short training courses for regulated healthcare professionals in Western medical acupuncture.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.