Research Review By Dr. Ceara Higgins©


Download MP3

Date Posted:

May 2018

Study Title:

The persistence of the effects of acupuncture after a course of treatment: a meta-analysis of patients with chronic pain


MacPherson H, Vertosick EA, Foster NE, et al.

Author's Affiliations:

University of York, United Kingdom; Memorial Sloan-Kettering Cancer Center, New York, USA; Keele University, United Kingdom; University of Southampton, United Kingdom; Technische Universitat Munchen, Germany; Group Health Research Institute, Seattle, USA; University of Zurich, Switzerland; Universitatsmedzin, Berlin, Germany; University of Maryland School of Medicine, Baltimore, USA.

Publication Information:

Pain 2017; 158: 784-793.

Background Information:

Patients with chronic pain may find the promise of any period of relief enticing, but more relevant is whether a given intervention can provide benefit over the longer term. The duration of the effect of an intervention, like acupuncture, has clear implications for health economics, as an intervention able to provide lasting results is more economically sound than one providing only short-term gains, in theory.

The authors of this study analyzed individual patient data from the Acupuncture Trialists’ Collaboration to establish the time course of acupuncture effects. Many trials of acupuncture were found to have measured outcomes at more than one point in time after the conclusion of treatment, allowing the authors to compare how the differences between groups changed between two time points post-treatment and attempt to estimate how much the effects of acupuncture treatment persist.

Pertinent Results:

The 29 trials included involved nearly 18000 subjects (longer-term data was available in 20 trials including 6376 subjects). Most trials included 8-15 treatments delivered over 10-12 weeks, with one trial having a 26-week treatment duration (1). In trials where acupuncture was compared to no acupuncture, the average number of treatments was 8 over 8 weeks, while for trials comparing acupuncture to sham acupuncture, the average number of treatments in both the acupuncture and sham groups was 10 treatments over 6 weeks.

Acupuncture compared with no acupuncture controls:
8 trials with a total of 2985 patients were included in this category. The fixed-effects estimate for the between-group comparison showed a non-significant decrease in the effect size of acupuncture after the end of treatment, with the decrease measuring approximately 9% in the acupuncture group at 12 months’ post-treatment. The differences in pain change scores between the acupuncture and no acupuncture groups remained relatively consistent at about 0.5 SD (standardized change score) (4) from randomization up to 1 year after treatment.

Acupuncture compared with sham acupuncture controls:
16 trials with a total of 4534 patients were included in this category. In these trials, a significant difference in pain was seen favoring the acupuncture group at the end of treatment. However, a significant decrease in effect size of about 50% was seen in the acupuncture group at 12 months post-treatment. Three neck pain trials (2, 3, 5) were included in the analysis that all showed decreases in the effects of acupuncture over time when compared with sham. A sensitivity analysis was run that excluded the neck pain trials, which showed a smaller nonsignificant reduction in differences in pain between groups over time. With the neck pain trials excluded, the decrease in effect of acupuncture over time reduced to 28% at 1 year. This suggests that the majority of the effects of acupuncture may persist for non-neck-related chronic pain. Overall, the benefits of both acupuncture and sham acupuncture seem to be largely continuous over time, up to 1 year after randomization.

When comparing average differences between acupuncture and the different control groups, no heterogeneity in how the effects of acupuncture decrease between sham and no acupuncture-controlled trials was found. This analysis included trials regardless of whether they showed differences between acupuncture and controls, which may help explain why the analysis did not find significant decreases in the effects of acupuncture over time. To account for this, the authors repeated their analysis excluding trials showing no significant effect of acupuncture when compared to sham or no acupuncture. This left the authors with 5 no acupuncture controlled trials, including 2059 patients, showing a non-significant increase in the effects of acupuncture compared to no acupuncture over time. 7 sham-controlled trials, including 1450 patients, showed a significant decrease in the effects of acupuncture compared to sham for every 3 months of follow-up. When a further exclusion of neck pain trials was undertaken, the remaining 5 trials, including 1203 patients, the decrease in the effect of acupuncture compared with sham was no longer significant.

Clinical Application & Conclusions:

These authors analyzed almost 30 trials involving almost 18000 subjects with chronic pain. Their results suggest the effects of acupuncture for chronic pain do not seem to decrease significantly over a 12-month period when compared to no acupuncture. An estimated 90% of the benefits of acupuncture when compared to controls will be sustained at 12 months. When using the upper bound of the confidence interval, about 70% of the benefit will be sustained.

The results were similar for acupuncture versus sham treatment when neck pain studies were excluded. However, it was clear that the effects of acupuncture for chronic neck pain do decrease over time when compared to sham acupuncture. This may be a result of the shorter treatment plans provided in the neck pain trials (2, 3, 5), which ranged from 3-4 weeks rather than the more common treatment plans lasting 6-8 weeks, or longer, featured in the other studies.

The major clinical implication of this study was that we can reassure chronic pain patients who are considering acupuncture that treatment benefits do persist beyond the end of treatment (with the possible exception of neck pain). This also has implications on cost-effectiveness. Many cost-effectiveness studies on acupuncture use the end of treatment as the end point for clinical effectiveness, essentially assuming that the benefits completely disappear as soon as treatment ends. When we change this end-point based upon the data showing effects for 12-months post-treatment, this dramatically impacts cost-effectiveness.

Study Methods:

A systematic literature review was performed on studies meeting the following inclusion criteria:
  • Trials of acupuncture for chronic pain
  • Trials where allocation concealment was determined to be adequate
  • Patients involved in the studies has nonspecific low back or neck pain, shoulder pain, chronic headache/migraine, or osteoarthritis
  • Trials compared acupuncture to a control group (including no treatment, wait-list, rescue medication, usual care or protocol-guided care, and sham acupuncture)
  • The primary outcome (pain) must have been measured at least twice after the end of treatment. In trials where control group patients were offered acupuncture treatment after the end of the initial treatment period, data from all time points after the control group began receiving treatment was dropped from the analysis.
Exclusion Criteria:
  • Trials where only one measurement was taken after the end of treatment
  • Trials where all outcome measures were taken only during treatment
  • Trials where the primary outcome was only measured after the control group began to receive treatment
The primary outcome measure for this meta-analysis was pain. As different studies used different measurements of pain, the authors attempted to standardize by dividing by the pooled standard deviation (SD) and multiplying by 100. Initial analysis did not show evidence that the effects of acupuncture differed by clinical indication/condition treated. Thus, the data was initially all pooled together. Pain intensity was used as the dependent variable and baseline score, time, and treatment group and an interaction term for group and time were used as predictors. Time was defined as the number of days since the end of treatment.

Study Strengths / Weaknesses:

  • A meta-analysis was performed drawing on individual patient data from high-quality randomized controlled trials for chronic pain. This resulted in the evaluation of a large data set of nearly 18,000 patients. This allowed the authors to explore subgroups in a way that is impossible when working with data from smaller trials.
  • Not all trials provided data from more than one post-treatment follow-up.
  • Only 8 of the 20 trials followed patients for 40 weeks or more.
  • Data from trials with longer and shorter follow-ups were used to estimate the effects of time on treatment. Therefore, just because the study showed no important decrease in the effect at 12 months, does not mean we can assume the effects will persist well beyond 12 months.

Additional References:

  1. Berman BM, Lao L, Langenberg P, et al. Effectiveness of acupuncture as adjunctive therapy in osteoarthritis of the knee: a randomized, controlled trail. Ann Intern Med 2004; 141: 901-910.
  2. Irnich D, Behrens N, Molzen H, et al. Randomised trial of acupuncture compared with conventional massage and “sham” maser acupuncture for treatment of chronic neck pain. BMJ 2001; 322: 1574-1578.
  3. Vas J, Perea_Milla E, Mendez C, et al. Efficacy and safety of acupuncture for chronic uncomplicated neck pain: a randomized controlled study. PAIN 2006; 126: 245-255.
  4. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture Trialists’ Collaboration. Acupuncture for chronic pain: individuals patient data meta-analysis. Arch Intern Med 2012; 172: 1444-1453.
  5. White P, Lewith G, Prescott P, et al. Acupuncture versus placebo for the treatment of chronic mechanical neck pain: a randomized controlled trial. Ann Intern Med 2004; 141: 911-919.