There has been an ongoing debate about placebos on SBM, both in the articles and in the comments. What does it mean that a treatment has been shown to be “no better than placebo?” If our goal is for patients to feel better and they feel better with placebos, why not prescribe them? Do placebos actually doanything useful? What can science tell us about why a patient might report diminished pain after taking an inert sugar pill? The subject is complex and prone to misconceptions. A recent podcast interview offers a breakthrough in understanding.
On her Brain Science Podcast Dr. Ginger Campbell interviewed Dr. Fabrizio Benedetti, a physician and clinical neurophysiologist who is one of the world’s leading researchers on the neurobiology of placebos. A transcript of the interview [PDF] is available on her website for those who prefer reading to listening. The information Dr. Benedetti presents and the expanded remarks by Dr. Campbell after the interview go a long way towards explaining the placebo phenomenon and its consequences for clinical medicine. Dr. Campbell also includes a handy list of references. I’ll try to provide a summary of the main points, but I recommend reading or listening to the original.
A common misconception is that the response to placebos is a purely subjective psychological response involving only the cortical level of the brain; but evidence is accumulating that real, measurable, objective subcortical neurophysiologic phenomena are involved. One of the first hints was a 1978 study showing that the placebo response to pain could be blocked by naloxone, a narcotic antagonist drug, indicating that the placebo must have actually caused an increase in endogenous opioids.
Placebos are used in clinical trials methodology, where a placebo group is compared to an active treatment group. Typically, about a third of subjects in the placebo group show improvement, but this doesn’t mean all of them have responded to the placebo. There are many other factors that can cause improvement after administration of any treatment, whether effective drug or placebo, including spontaneous remission, regression to the mean, and various kinds of bias on the part of patients and experimenters.
Clinical trials are done to look for a difference between the outcomes with placebo and the outcomes with the drug. Researchers in those trials are not interested in trying to understand why there is an improvement in the placebo group. That requires a different kind of study. Dr. Benedetti is using “placebo balanced design” to tease out the influence of verbal suggestions — expectations — on the action of drugs. Subjects are divided into four groups. The first group of subjects receives the active treatment and is told it is the active treatment (the truth). The second group receives the active treatment and is told it is placebo (a lie). The third group receives placebo and is told it is the active treatment (a lie). The fourth group receives placebo and is told it is placebo (the truth). This design allows researchers to separate the specific effects of the active treatment itself from the nonspecific effects of suggestion and expectation.
In studies of surreptitious vs. open IV morphine administration, patients experienced less pain relief from the morphine when they were unaware that they were getting it. In one brain imaging study, the metabolic response to methylphenidate decreased by 50% when subjects were told they were getting a placebo.
The placebo response is not limited to pain. In Parkinson’s disease, placebos have been shown to cause a 200% increase in dopamine release in the brain, with an alteration in the activity of neurons and a corresponding clinical improvement in motor function. Dr. Benedetti has even documented a physiologic placebo response at the level of single neurons. In deep brain stimulation studies, the stimulator doesn’t work as well if patients are told it is off when it is actually on.
There is not just one “placebo effect,” but many different placebo effects that work by different mechanisms, including (1) anxiety reduction, (2) activation of the reward mechanism (with dopamine release in the nucleus accumbens), and (3) learning.
Placebo responses can be divided into two types: conscious and unconscious. Conscious responses involve suggestion and expectation. Unconscious responses occur with classical Pavlovian conditioning. If you give morphine 3 days in a row and on the 4th day you replace it with a placebo, almost 100% of patients will respond to the placebo as if they had received morphine. When a pill is administered, there is a completely unconscious association between two stimuli: the pharmacologic effect and the psychosocial context, which may include such things as the color and shape of the pill. Unconscious conditioning has been demonstrated in both animals and humans.
Conscious expectations are important for conscious physiological functions like pain and motor performance. Unconscious conditioning is more important for unconscious physiological functions like hormone secretion and activation of different immune mediators. Dr. Benedetti gives this example: if you give a patient a placebo and tell him it will increase his growth hormone levels, nothing happens. But if you give a drug that really does raise GH levels for two days and on the third day replace it with a placebo, the levels will rise with the placebo.
Dr. Benedetti gives placebos to subjects in a research setting; he doesn’t recommend giving them to patients in the doctor’s office. Placebo studies are important for understanding how the brain works, but Dr. Benedetti is not sure that they will lead to any clinical application. He says
the take-home message for clinicians, for physicians, for all health professionals is that their words, behaviors, attitudes are very important, and move a lot of molecules in the patient’s brain. So, what they say, what they do in routine clinical practice is very, very important, because the brain of the patient changes sometimes… there is a reduction in anxiety; but we know that there is a real change…in the patient’s brain which is due to… the ‘ritual of the therapeutic act.’
Experimentally, morphine requirements can be reduced by starting with morphine, substituting a placebo for later doses and periodically reinforcing the conditioning with morphine. This sounds like a good thing, but before it could become a useful option in clinical practice, it would have to overcome a number of ethical and practical hurdles. The unconscious placebo responses require conditioning and the conscious ones are problematic too. Even if some patients might get a degree of pain relief from just being handed a sugar pill with a strong suggestion, the effects can’t compete with effective pain treatments: the response is generally smaller, less reliable, less predictable, and not sustainable over the long term. I can’t imagine that ever becoming standard medical practice, and not just because of the ethical issues.
The doctor-patient relationship is essential to medicine, and placebo effects are an inherent part of that interaction. Communication is vital. Doctors must tell patients what a treatment is supposed to do. When pain medication is given without the patient’s knowledge, it doesn’t work as well. A clear understanding of the diagnosis is important: a positive diagnosis of something menacing like cancer can make the patient anxious and more aware of symptoms, while a negative diagnosis can reassure, relieve anxiety, and divert attention elsewhere. Lying to patients would undermine a trusting doctor-patient relationship and ultimately even interfere with the ability to evoke placebo responses. Prescribing placebos is uniformly rejected by medical ethicists: instead, we can put our increasing knowledge of placebo neurophysiology to good use without lying to patients.
This article was originally published in the Science-Based Medicine Blog