Sunday, January 1, 2012

Acupuncture a Dud for Pain, But Evidence is Weak

Acupuncture In a multicenter clinical trial, acupuncture proved to be disappointing for the relief of musculoskeletal pain in women taking aromatase inhibitors for breast cancer. Consequently, after an interim analysis, the trial was stopped early. However, this study provides lessons in the complexities of pain research on acupuncture and how weaknesses of evidence preclude valid conclusions.

Research teams in the United States and United Kingdom led by Ting Bao, MD, of the University of Maryland Greenebaum Cancer Center, enrolled 47 postmenopausal women in a double-blind, randomized, sham-controlled trial to assess beneficial effects of acupuncture. All subjects had early-stage breast cancer and experienced painful musculoskeletal symptoms related to aromatase inhibitor therapy for their cancer; none of them had received acupuncture in the prior year. The acupuncture group (N=23) received 8 weekly sessions targeting 15 acupuncture points, while the control group (N=24) received a sham acupuncture procedure at 14 non-acupuncture points. The sham therapy utilized the Park Sham Device, consisting of a nonpenetrating, flat-tipped, retractable acupuncture needle in a telescoping tube [more info here].

Results were reported in a poster session at the 2011 San Antonio Breast Cancer Symposium [Bao et al. 2011] and normally would receive little recognition; however, this study also was featured more conspicuously in an edition of Internal Medicine News [12/16/2011 here]. The researchers note that both real and sham acupuncture produced reductions in the Health Assessment Questionnaire Disability Index [HAQ-DI described in PDF here], which was the primary study end point. Starting from a median baseline HAQ-DI score of 0.75, 35% (8/23) of patients in the real acupuncture group achieved a minimum clinically important improvement — predefined as a 0.22-point reduction in HAQ-DI — as did 54% (13/24) of women in the sham acupuncture group.

Both groups also exhibited improved scores on a pain VAS (visual analog scale) after 8 weekly real or sham treatment sessions. From a baseline median of 50 mm (out of a possible 100 mm) the median score in the real acupuncture group was reduced to 41 mm (difference= 9 mm), and to 36 mm in the sham group (difference = 14 mm).

Acupuncture was relatively safe, without significant side effects reported in either study arm. However, Bao and colleagues noted that more than half (55%) of subjects in each group accurately guessed their treatment assignment to either real or sham acupuncture.

The researchers had originally intended to enroll 100 patients in the trial; however, it was halted early based on the interim results reported above. Overall, the researchers concluded that there was no advantage demonstrated by real acupuncture compared with the sham procedure in ameliorating musculoskeletal symptoms in this patient population.

COMMENTARY: Finding safe and effective pain therapies for this population of women with early-stage breast cancer is important because half of patients on adjuvant aromatase inhibitors experience musculoskeletal symptoms severe enough to incur poor medication compliance or even treatment discontinuation. In the past, analgesic acupuncture as compared with placebo/sham procedures has demonstrated very mixed results in trials for various painful conditions and there sometimes are confounding factors muddling outcomes, especially placebo effects [see prior UPDATES on this topic here].

Similarly, there are a number of concerning aspects of this present study, which was funded by a Komen Foundation grant and an American Society of Clinical Oncology Young Investigator Award:

  • Reporting of median (middle) values on outcome measures, as the authors did, is unhelpful without also indicating the ranges or Standard Deviations to understand the spread of scores. And, in most cases, one is more interested in mean values to assess outcomes; that is, how “average” patients would fare, even though means can be misleading at times if there is a high dispersion of scores.

  • For example, the median baseline VAS of 50 in this study, indicating a moderate pain level, does not denote the range of pain scores in this population, which might have been diverse. Therefore, it cannot be determined if patients with either more or less severe pain at the outset might have responded better to acupuncture.

  • Furthermore, age makes an exponential difference in HAQ-DI scores, which can range from 0 (no functional disability) to 3 (severe disability). Normative population scores for women aged >50 years typically range from (mean±SD) 0.16±0.37 (age 50-54) to 0.77±0.72 (age 75-79).

    Therefore, depending on the age distribution of women in the study by Bao et al., which is unstated, the 0.75 median score at baseline may or may not represent an excessively high or atypical degree of functional disability at the outset. The researchers do concede that the HAQ-DI might not have been a sufficiently sensitive measure to detect differences between real and sham acupuncture groups.

  • Still, more than a third and a half of subjects in the real and sham acupuncture groups, respectively, did achieve a minimally significant 0.22 reduction in HAQ-DI scores. And, both groups also exhibited small to moderate effect sizes for reductions in pain scores on the VAS.

  • Therefore, there were trends toward improvement due to the interventions; albeit, as has been found in other research on acupuncture, sham/placebo approaches also demonstrate some efficacy. The researchers acknowledge that the sham acupuncture in their study might not have been an inert procedure; in fact, from the interim data, it could be that the Park Sham Device has better analgesic potential than real acupuncture. Of course, the results are confounded by the fact that more than half of subjects, supposedly blinded to treatment condition, suspected correctly whether or not they were receiving real acupuncture.

  • Perhaps of greatest consequence, with only 47 total subjects, the study was underpowered to distinguish even large effect sizes as being statistically significant [see effect-size discussion in UPDATE here]. We have previously discussed how stopping underpowered trials early raises questions about whether the study should have been undertaken in the first place and provides unreliable evidence of either success or failure, since a larger or fully enrolled trial might have demonstrated much more definitive results [see discussion of statistical power here].

Therefore, studies such as this — briefly described in a conference poster or media report — usually do not offer reliable information of any value for understanding the analgesic potential of acupuncture. Unfortunately, however, this trial may be inappropriately cited in the literature as demonstrating the failure of acupuncture when, in actuality, it portrays weak results possibly due to inadequate study design and follow-through.

Discrediting any therapy on the basis of poor- or low-quality research evidence is not good science. Yet, it is now unlikely that a larger, more definitive trial of acupuncture in this population of women will ever be attempted in the future.

REFERENCE: Bao T, Tarpinian K, Medeiros M, et al. A Multi-Center Randomized Controlled Double Blind Trial Assessing the Effect of Acupuncture in Reducing Musculoskeletal Symptoms in Breast Cancer Patients Taking Aromatase Inhibitors: First Interim Analysis. Presentation at the San Antonio Breast Cancer Symposium, December 6-10, 2011, San Antonio, Texas. Poster: P4-12-13.