Effects of acupuncture in reducing attrition and mortality in HIV-infected men with peripheral neuropathy

A clinical trial reported in JAMA (Shay et al, 1998), involving acupuncture and amitriptyline in HIV-infected patients, concluded that there was no effect for either acupuncture or amitriptyline on neuropathic pain. However, a recent reassessment of this study showed that there were really three different and independent clinical trials, each with a different research design, which had been combined into a single database and consequently analyzed with a relatively insensitive statistics. When only the first substudy, factorially crossed design involving acupuncture and amitriptyline, was reanalyzed by itself using more powerful statistics, it was found that acupuncture and amitriptyline both worked independently to reduce pain, but also that acupuncture worked best in the absence of amitriptyline, and that there may have been adverse events associated with the combination of the two treatments. The present study reports the reanalysis of the second of the original independent studies involving only acupuncture and sham acupuncture, to determine whether the results confirm acupuncture-related findings from the first substudy.

Subjects were 114 HIV-infected men with pain associated with peripheral neuropathy in the early 1990s, when antiretroviral drug cocktails were just beginning to be available in experimental form.

The second of the independent studies in the original report by Shlay et al involved a single factor with two levels: a 14-week standardized acupuncture regimen and its control (off-point sham acupuncture). In addition, physical functioning at baseline (high or low, based on the Karnofsky scale), was factorially crossed with the acupuncture factor in our analyses. Primary data were reanalyzed using repeated-measures ANCOVA in an intention-to-treat procedure, and categorical data were analyzed by the Pearson chi-square test.

Pain intensity, pain relief, mortality, and attrition.

Whereas the results were inconclusive for the pain measures, acupuncture had a strong and positive effect on attrition and mortality. These results were most pronounced among patients with poorest physical functioning at the beginning of the study. Overall, acupuncture was associated with lower attrition rate (27.6% vs. 44.6%, P = .058), and a zero mortality rate (0% vs. 12.5%, P = .047). This protective effect of acupuncture was visible primarily in subjects in poorer health (0% vs. 23.8%, P = .047).

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). Moreover, these results replicated most of the findings that did not involve the presence of amitriptyline from the initial independent study in this research project. The combined results of these two studies strongly support the importance of recognizing that interactions involving acupuncture and other treatments, may positively as well as negatively modify main effect results in clinical trials, and thus must be recognized and systematically explored. Findings are discussed in terms of their implications for moving toward a whole-systems approach to biomedical research.