I have written about the placebo effect before, but a recent update to a Cochrane review entitled Placebo Interventions for All Conditions has prompted me to revisit the subject. It has been clear for some time that the so-called placebo effect is not an example of the power of the mind to overcome disease. There is no evidence that therapies which have no measurable biological affect can truly alter the progression or outcome of an illness. No matter how you feel when given a deliberate placebo or an inert CAM therapy, you don’t live longer, your tumor doesn’t get smaller, you blood sugar doesn’t become normal, and the objective effects of your disease on your body don’t change.
What has seemed reasonable to me in the past is the idea that there might be real benefits in terms of subjective symptoms, such as pain or nausea, attributable to placebo effects. However, I am a little less confident of that in light of this recent review.
The review was a metanalysis of clinical trials which included both placebo and no treatment groups. As expected, no significant clinical effects were found for any objectively measurable variable. However, even in the case of subjective variables like pain and nausea, the best the authors could say was, “in certain settings placebo interventions can influence patient-reported outcomes, especially pain and nausea, though it is difficult to distinguish patient-reported effects of placebo from biased reporting.”
For pain, some trials reported little or no effect, and those that reported a large effect were all conducted by the same group of researchers, which raises suspicions about the reliability of the findings. Findings for most other subjective measures either showed no benefit from placebo, or effects that varied enormously from one study to another, suggesting differences in the studies were responsible for differences in the findings. If the placebo effect were a consistent, reliable phenomenon one would expect it to show up about the same regardless of who is studying it.
The metanalysis also found that how much benefit patients seemed to get from placebo treatments depended heavily on whether they thought they might be getting a placebo treatment. Inert treatments worked better in trials where the subjects were told that no placebo would be given.
These findings and others like them make it seem very likely that the placebo effect is less a feature of the human mind than it is an artifact of clinical trial design, a conglomeration of biases and measurement errors rather than mind over matter. The better one controls for biased reporting of effects, the less effect one sees. This is not how the results look when one studies an unequivocally effective therapy, even one involving subjective symptoms, such as NSAIDs for arthritis pain. Effects from real therapies are consistently seen in different studies by different researchers, and they do not disappear when the trials become better controlled for bias.
Of course, if someone tells you they have pain or nausea, and then you give them a placebo and they say they have less pain or nausea, can you really say the effect isn’t real? If these phenomena are ultimately subjective, than regardless of the biological events underlying them the perceptions of the sufferer must be the critical determinant of benefit. Still, most of the research on placebos and sham therapies suggests that when well-designed studies are conducted, these benefits are small and do not reach anything like the level of clinical significance of true therapies for these symptoms. And, as I have discussed before, taking advantage of what benefit may occur due to placebo effects requires successfully deceiving the patient, which raises significant ethical concerns. The lesson of this review seems to be that placebo effects may turn out to be mostly illusory, artifacts of inadequate study design, and we should be hesitant to rely on them for treatment of any clinically significant symptoms.