There is evidence that the magnitude of the placebo effect produced by a device is greater than that produced by a pill. If this is the case, it has significant ramifications for all trials involving devices and for our understanding of the role of the placebo effect in randomized controlled trials (RCT). This two phase study 1)investigates the role of the placebo effect in RCT's and 2)conducts two trials of treatments for persistent upper extremity pain secondary to repetitive strain injury (RSI), including carpel tunnel syndrome. In Phase I. 240 patients with RSI are randomly assigned to receive a placebo device (a recently validated sham acupuncture device) or a placebo pill (dummy amitriptyline). Our primary hypothesis is that patients will respond better to the sham device than the placebo pill. A finding that sham acupuncture produces a greater placebo response than a placebo pill has important implications for the interpretation of results in trials that compare devices to sham devices, devices to pills, and medical management to surgery. Phase II randomly assigns patients from the sham acupuncture arm of Phase I to receive either TCA or continue to receive the sham version.

Patients in the placebo pill arm of Phase I will be randomly assigned to receive either AMI or continue receiving the placebo pill. From the patients'perspective, the shift in treatment assignment from Phase I to II should not be noticeable. Phase II will allow us to test whether the active treatments outperform their respective placebos. Both of these treatments have shown promise in small studies, but neither has been prospectively studied in a large trial with appropriate controls. Because Phase I also functions as a run-in period for Phase II, analysis combining both phases will allow us to examine whether a run-in has methodological advantages in a device trial. Moreover, combined analyses permit testing whether patients level of response to placebo in Phase I affects their response to active treatment Phase II. A positive finding here would contribute importantly to our understanding of the role of the placebo in RCTs.