Clinical articleOriginal researchEffects of Acupuncture in Reducing Attrition and Mortality in HIV-Infected Men with Peripheral Neuropathy
Introduction
Acupuncture has been shown in a number of studies to be effective for reducing pain involved in a variety of different medical conditions, a fact that has been acknowledged by the National Institutes of Health.1 Despite the abundance of positive findings supporting the relationship between acupuncture and reduced pain, some studies do not show a clear-cut benefit to acupuncture, showing either weak results or no effect.1, 2, 3 Vickers4 recently demonstrated that the weak findings in four acupuncture pain studies5, 6, 7, 8 were due in part to inadequate statistical analysis. However, when a more sensitive statistical method (analysis of covariance, ANCOVA) was applied to the raw data from these four studies, stronger results emerged in support of the effectiveness of acupuncture.
Another study of acupuncture for pain, involving peripheral neuropathy in patients with advanced HIV disease, concluded that neither acupuncture nor an active comparison treatment, amitriptyline, had any effect on pain intensity.3 This study was a large, multisite clinical trial that is unique in that it involved severely ill patients with outcomes that include pain management as well as attrition and an unusually high death rate, and is unlikely to ever be replicated due to prohibitive costs and the nature of the illness: advanced HIV disease at a time when adequate treatment was not yet available.
As shown in Figure 1, reprinted from Shlay et al, that study actually consisted of three independent studies or “options.” The Factorial Option was their originally planned study: an elegant 2 × 2 factorially crossed design involving acupuncture and amitriptyline and their corresponding control groups, sham acupuncture, and placebo pill. Options 2 and 3 were added two years after the study began, and involved one-way designs involving only acupuncture and its sham condition (in Option 2), or amitriptyline and its placebo pill condition (in Option 3).
Because these latter two designs were not compatible with the 2 × 2 design of the “Factorial” Option, the originally planned statistical analyses were substantially modified to a point where it was no longer possible to directly compare amitriptyline to acupuncture. The statistical design that was actually used is presented in the legend of their Figure (2), reproduced here as Figure 1, ended up being much less sensitive than their originally planned 2 × 2 factorial analysis of variance (ANOVA).3 Their modified statistical approach assumed that there was no interaction between acupuncture and amitriptyline, because, as shown in the legend of Figure 1, components of each factor were included in the other factor. In other words, the two factors in the 2 × 2 design, acupuncture and amitriptyline, were not independent of each other, thereby making direct comparisons between acupuncture and amitriptyline impossible. For all these reasons, it seemed likely that the results reported by Shlay et al substantially underestimated the true effects.
To examine this possibility, Shiflett and Schwartz 9 reanalyzed the original raw data from that study using only the patients enrolled in their original 2 × 2 factorial design option, to determine whether the increased statistical power achieved by using this approach, despite a reduced sample size, would yield more positive results for either acupuncture or amitriptyline. In contrast to the findings reported by Shlay et al, Shiflett and Schwartz found that both acupuncture and amitriptyline independently aided in pain relief, and perhaps more importantly, acupuncture was associated with fewer deaths and a much lower attrition rate than in any other condition, except in the presence of amitriptyline, where there was a significant increase in the risk of death and attrition. These effects were most pronounced in patients in poorer health at baseline.
The purpose of the present study was to reanalyze the data from the “Acupuncture” Option (Option 2, see Figure 1), which consisted of an independent study involving only the acupuncture-sham acupuncture factor (ie, the amitriptyline option was not included in this study), but which was otherwise identical to the Factorial Option in terms of research design, subject pool, randomization, blinding, and outcome measures, in order to determine whether these data would confirm the equivalent findings from the Factorial Option. The primary question was whether the findings of reduced attrition and mortality with acupuncture, especially in patients in poorer health at baseline, would be replicated in the second experiment.
Section snippets
Methods
Based on information graciously provided to us by the senior author of the original report (Dr. Judith Shlay), we obtained their original database from the NTIS, where the data had been archived in accordance with National Institutes of Heath and JAMA recommended procedures for sharing data. Information presented below, describing their methods, is essentially a summarized version of the information presented in their article originally published in JAMA.3
Pain Relief
There was a moderate tendency for pain relief to occur over the first six-week period, regardless of acupuncture condition, and then to stabilize at that level through the 14th week. Initial health status tended to be associated with general pain relief over time, with those in poorer health experiencing less relief than those in better health (P = .072). Within this context, acupuncture was more effective than sham at relieving pain (P = .015). Most of the relief for both conditions had
Discussion
The pattern of findings observed here generally replicate the positive findings reported in Shiflett and Schwartz9 concerning the effects of acupuncture in reducing attrition and mortality in HIV-infected men with peripheral neuropathy. There was strong evidence of the protective effects of acupuncture on mortality and attrition in similarly infected patients, particularly in those who were experiencing poor functioning at baseline, thus confirming results from the first study of this series.9
Conclusion
Acupuncture was clearly effective in reducing attrition and mortality in this sample, especially when health status was taken into account, but results for pain relief were mixed. These results add further evidence that the use of the most sensitive statistics available increases the chance of detecting actual effects due to acupuncture (and other treatments as well). These results also confirm most of the findings that did not involve the presence of amitriptyline from the initial independent
References (20)
- et al.
Randomized clinical trial comparing the effects of acupuncture and a newly designed placebo needle in rotator cuff tendonitis
Pain
(1999) - et al.
A randomized comparative trial of acupuncture versus transcutaneous electrical nerve stimulation for chronic back pain in the elderly
Pain
(1999) - et al.
Pressure pain threshold and needle acupuncture in chronic tension-type headache—a double blind placebo-controlled study
Pain
(2000) - et al.
Statistical reanalysis of a randomized trial of acupuncture for pain reveals positive effects as well as adverse treatment interactions on pain, attrition and mortality
Explore (NY)
(2010) - et al.
Ratio scale of sensory and affective verbal pain descriptions
Pain
(1978) Acupuncture
JAMA
(1998)- et al.
Acupuncture treatment for pain: systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no acupuncture groups
BMJ
(2009) - et al.
Acupuncture and amitriptyline for pain due to HIV-related peripheral neuropathy: a randomized controlled trial
JAMA
(1998) Statistical reanalysis of four recent randomized trials of acupuncture for pain using analysis of covariance
Clin J Pain
(2004)- et al.
Randomized trial of acupuncture compared with conventional massage and “sham” laser acupuncture for treatment of chronic neck pain
BMJ
(2001)
Cited by (16)
Acupuncture for HIV/AIDs and its complications: A systematic review and meta-analysis
2024, European Journal of Integrative MedicineNeuromuscular complications of HIV infection
2018, Handbook of Clinical NeurologyCitation Excerpt :A randomized trial comparing acupuncture to amitriptyline and placebo found no significant reduction in pain when compared to placebo (Shlay et al., 1998). In a reanalysis of the individual data, different authors argued that acupuncture and amitriptyline were both effective at reducing pain, although acupuncture worked best in the absence of amitriptyline (Shiflett and Schwartz, 2011). A subsequent study of acupuncture and moxibustion (burning of mugwort leaf) demonstrated reduction in neuropathy when compared to a sham intervention (Anastasi et al., 2013).
Peripheral neuropathy in HIV: An analysis of evidence-based approaches
2014, Journal of the Association of Nurses in AIDS CareCitation Excerpt :Shlay et al. (1998), in an early clinical trial examining the role of acupuncture and amitriptyline in HIV-infected patients, concluded that these methods had no effect on neuropathic pain. However, Shiflett and Schwartz (2011) suggested that acupuncture was effective in reducing attrition and mortality in this sample, especially when health status was taken into account, but results for pain relief were mixed. Anastasi, Capili, McMahon, and Scully (2013) investigated the role of acupuncture/moxibustion (acu/moxa) for HIV-SN in a randomized controlled pilot study to investigate whether acu/moxa significantly reduced neuropathic pain.
Dolor neuropático y acupuntura. Evidencia científica de su efectividad
2013, Revista Internacional de AcupunturaEvidence-Based Pain Guidelines in HIV Care
2013, Journal of the Association of Nurses in AIDS CareCitation Excerpt :The evidence for the effectiveness of acupuncture for pain management is inconsistent. There is inconclusive evidence for acupuncture as a treatment for the pain of HIV-associated neuropathy (Shiflett & Schwartz, 2011). And in two Cochrane reviews, insufficient evidence was found to support the use of acupuncture for pain in persons with cancer (Paley, Johnson, Tashani, & Baynall, 2011) or low back pain (Standaert et al., 2011).