A new meta-analysis of over 1,200 drug trials has found that 49% of participants given placebos reported some type of medical problem during the experiment.
The research, published in Trials journal, looked at randomised drug trials that included over 250,000 participants on placebos and found that not only were these participants reporting "adverse" medical events, they dropped out of trials at a higher rate than groups that were entirely untreated.
On average, 5% of participants taking placebos dropped out from the drug trials due to adverse medical events.
The authors of the paper state that because these adverse reactions can't be directly caused by a substance that has no therapeutic effect, the dropout rates could be caused by participants with an underlying medical condition misattributing symptoms such as headaches to the placebo.
However, the researchers indicate that these adverse medical events could also be caused by a consistent "nocebo" effect in drug trials.
There's evidence for this effect in one of the studies observed in the analysis: due to a difference in hospital administration, half of one placebo group received warnings of possible gastrointestinal side effects, while the other half did not.
Receiving the warning resulted in a six-fold increase in the number of participants withdrawing from the study due to gastrointestinal symptoms.
The nocebo effect can be thought of as the direct opposite of the placebo effect, in that it worsens patients' conditions rather than improving them.
The placebo effect first began to be observed after World War II when randomised placebo-controlled drug trials became the primary method of testing pharmaceuticals.
At the most basic level, this means that standard drug trials involved a randomly-assigned group of people taking the chemically-active drug, while the other group were given sugar pills. The results were then compared.
Scientists began to notice that the patients who were being treated with the placebo sometimes improved markedly.
American anesthesiologist Henry Beecher reported in 1955 that approximately 35% of patients with severe postoperative pain can be effectively treated with placebos.
The placebo effect has been explained by several strong psychological factors including confirmation bias (if a patient believes they are going to improve, they will ignore their continuing or worsening symptoms) and expectation.
One 2007 paper states that placebo "effects are treatment effects caused not by the physical properties of a treatment but by the meaning ascribed to it".
Expectation can be manipulated by the level of medical intervention that a patient experiences. For example, use of fake acupuncture needles has been shown to be more effective than taking placebo pills in treating muscle pain.
One literature review found that 74% of patients who underwent a placebo surgery (a skin incision without a true operation) for arm issues showed improvement in their problematic arm. In over half of the surgeries observed, the effect of the placebo was equivalent to that of the actual surgeries.
The physical mechanism of the placebo effect is thought to be driven primarily by a a release of the brain's natural opioids in the regions of the cortex critical for pain assessment and management.
Associate professor Ben Colagiuri, a placebo researcher from the University of Sydney, told BuzzFeed News that this release of natural opioids causes a "cascade of effects where this expectancy is actually changing our physiology".
The nocebo effect – a patient's expectation that they are going to experience a worsening of symptoms – has been observed in many different forms: it can exacerbate problems of nausea related to chemotherapy, increase experiences of pain, and has been shown to worsen symptoms of Parkinson's disease during treatment.
Colagiuri said that while the nocebo effect is less well investigated than the placebo effect, the biological mechanisms that cause expectation is similar.
"In the case of pain, if somebody expects pain it might trigger negative mood states, so more anxiety, and that anxiety can lead to the release of various neurotransmitters in the brain like catecholamines, and that can lead to the increase in pain reports," he said.
Colagiuri said that while the placebo and nocebo effects might be fascinating because they appear as though "something's coming from nothing", there are biological explanations for both.
The nocebo effect presents some important methodological problems for researchers conducting drug trials. While researchers have an ethical obligation to inform patients of side effects, the increased dropout rate and medical issues experienced by placebo-taking participants can impact on the validity of the trial.
Research is beginning to emerge on how this effect can be mediated by reversing warnings of side effects to change participants' expectations.
Colagiuri said that when patients are informed of the possibility of side effects, it will generally be framed as, for example, "30% of patients will experience headaches". Reframing this warning as a positive statement ("70% of patients will not experience headaches") could alter patient expectation.
One study led by the University of NSW published this year found that positive framing when patients are prescribed capsules of medication can reduce the nocebo side effects in the short term.
However Dr Jeremy Howick, lead author of the Trials meta-analysis, told the BioMed Central blog that these solutions are still under-researched and scientists are currently searching for the best methods for telling patients about the potential benefits or harms of drug trials.