Testing for Placebo Effects Using Data from Blinded, Randomized Controlled Trials

This paper proposes a test for the existence of placebo effects, as described by the so-called expectancy theory. This theory, which is the dominant medical theory of how placebo effects operate, posits that health outcomes rise in individuals' beliefs about the probability that they are getting a beneficial treatment and their beliefs about the efficacy of that treatment. Blinded, randomized, controlled trials provide near-perfect environments in which to test this theory because they offer objective, controlled manipulations of subjects' beliefs about treatment. If the expectancy theory is correct, outcomes in trials offering a higher probability of receiving an experimental treatment should be superior to outcomes in trials offering a lower probability of receiving that treatment, conditional on treatment assignment. The paper applies this test to data from over 200 trials of anti-ulcer medications and finds robust evidence of placebo effects in trials of H2-blockers (e.g., Zantac, Tagamet and Pepcid) and of proton-pump inhibitors (e.g., Prilosec, Nexium, and Prevacid). Indeed, trials of H2-blockers manifest placebo effects that are 50 percent as large as the physiological effects of these medications. Because placebo effects are not confined to clinical trials, this result suggests that the standard difference-in-means estimator of treatment effects may seriously underestimate the efficacy of anti-ulcer medications