Friday, December 29, 2017
Language and the Nocebo Effect
Arthur Barsky wrote an article in this week's JAMA entitled "The Iatrogenic Potential of the Physician's Words." Dr. Barsky is one of those authors I have been reading over the past 20 years. He is a psychiatrist and an expert on psychosomatic medicine. Like a number of authors, when I see his name - I just read the article. It is automatic. This article is brief but it contains a lot of wisdom about how communication with patients can affect the expected outcome of medical treatment. There are several points in the article that just jump out at you if you are an experienced clinician and have followed many patients.
The single most important sentence in the article starts the second paragraph:
"Somatic symptoms and underlying disease do not have a fixed, invariable, one-to-one equivalence." Physicians of course are trained to see a multitude of presentations of illness. Influenza is a good example. The majority of people are more ill than if they have the common cold, but a significant number are critically ill and dying. For example in 2016, the CDC estimated that about 24.6 million people got the flu, 308,000 were hospitalized, and 11,995 died. Theoretically there are a number of people with few or no symptoms who can still spread the virus. There are a significant number of factors in both the disease vector and the host that determine a wide array of outcomes. Infectious diseases provide the most straightforward example or variability in symptoms, underlying illnesses and outcome.
With the chronic complex, polygenic illnesses that require ongoing medical care the situation is less clear. In this case we have a patient who is constantly updating their assessment of their problems, symptoms, and state of their illness. They make similar determinations about any medicine or treatment that their physician wants them to take. A significant part of that determination is based on the information that they obtain from their physician or other sources and more importantly how it is presented. From a psychiatric perspective it should be presented as neutral as possible, but patients vary greatly in their reactions to even fairly dry information. As an example, I typically provide patients with MedlinePlus information on their medications. I consider these documents to present all of the essential unbiased information. It is still common to encounter people who tell me: "Don't give me that doctor! I know if I read it I will get every side effect that is listed!" I honor their request and reassure them that they can call me if there are any problems.
Dr. Barsky focuses on the informed consent aspect of treatment and how the information presented can affect the patients reaction to the treatment. He gives examples of men warned about the sexual side effects of beta blockers getting roughly twice as much sexual dysfunction and back pain patients undergoing spine imaging having more pain, functional impairment and physician visits. Psychiatry has the additional burden of many more vocal critics promoting the side effect burden of medications, often to the point that it seems like there was a conspiracy to "cover them up". That results in patients in the office complaining about what they have read online, even before the informed consent discussion begins.
There is a discussion of the nocebo effect or adverse effects to placebo treatment that does not receive enough discussion within medicine. I had early familiarity with this phenomenon from my work with clinical trials of pharmaceuticals (antidepressants, anxiolytics, antipsychotics). In each case the active drug was compared to placebo and patients were followed closely. In those days we were allowed to break the blind to tell patients if they were on active drug or placebo. The first time I encountered the nocebo effect, I was treating a young man with an experimental anxiety medication. He bitterly complained of many side effects initially and at the second or third visit said that he could no longer stand it. He needed to be taken off the medication because it was making him extremely ill. I broke the blind and told him that he was taking an inert placebo. He was somewhat embarrassed but the symptoms resolved during the time he was sitting in my office. Since then I have seen the nocebo effect hundreds of time in clinical practice. Many of the predisposed claim a high sensitivity or allergy to multiple medications and warn me ahead of time that they need to start on the lowest possible dose of medication. Even then it is dicey.
Over the years that led me to develop a uniform informed consent decision designed to minimize the nocebo effect. I think it is fairly straightforward to provide all of the relevant information in a manner that transmits both the side effect but also the likelihood of developing the side effect to the patient. Worst case scenarios are never avoided. For example, I always warn people on serotonergic medications about the risk of serotonin syndrome and the fact that it can be lethal. I also let them know that the chances of developing serotonin syndrome are very low. I use the same approach for tardive dyskinesia and other tardive syndromes, diabetes mellitus, cardiac arrhythmias, metabolic syndrome, seizures, neutropenia and agranulocytosis, and neuroleptic malignant syndrome. But also give them an estimate of how likely it is that they will develop that side effect or how often I have seen the side effect in 30 years of practice. That approach seems to have the expected effect on false positive reports of these symptoms and in fact I am much more likely to notice threshold symptoms (hypophonia, micrographia, etc) than the people who I warn about the side effects.
A second tier of symptoms surrounds the issues of sexual side effects and weight gain. A significant number of people are averse to trying any medication that might result in either side effect. That can occur even when they want to take the medication. In that situation it is useful to have a comprehensive reference that described the entire range of side effects in clinical trials for specific medications and present the facts. The facts will generally illustrate that the side effect of concern is not universal by any means and can be addressed if it happens in the case of sexual side effects or proactively (at least in some cases) in the case of increased appetite and weight gain.
Given the long list of potential side effects, reassurance that they will be assessed and a plan put in place to address them is the best approach to informed consent. The other discussion that most people find reassuring is that I have a low threshold for stopping a medication that causes any side effect that a patient finds distressing. During residency I had the experience of stopping an antidepressant medication that a patient had been on for 5 years. He was complaining primarily of a flu-like syndrome and headaches. After that medication was stopped those symptoms cleared up entirely. Since that incident, I routinely advice patients that I don't advise that they try to "get used to" medications and that I would prefer we just stop a medication if it is causing problems and try a different approach where possible.
Barsky invokes viscerosomatic amplification as a likely cause of information affecting symptom perception. He provides a very interesting supplement outlining a process whereby both benign but noticeable bodily sensations and disease symptoms of varying intensity are biased by information provided by the physician leading to a feedback loop of increasing anxiety, vigilance, and reactivity. Psychiatrists used to treating anxiety disorders are probably very familiar with this mechanism. I routinely ask those patients if they have cardiac awareness. Are they aware of their heart beating as they lay awake in bed at night waiting to fall asleep? Is it more intense than that - like their entire body is pulsating? What is their theory of what is happening? Are they concerned that there is something wrong with their heart? That phenomenon suggests that anxiety plays a significant role in selective attention and hypervigilance that is internally focused. It also suggests a major role for temperament and personality factors in the hypervigilance process since so many patients experience the anxiety at that level.
Concrete steps to minimize the symptom amplification process are discussed and they logically follow the proposed mechanism. The meaning to the patient of the symptoms needs to be explored. Education can be provided about both the nocebo effect and the amplification process. I typically use my research vignette for the nocebo discussion. The purported evolutionary aspects of anxiety and how it results in theories including somatic theories is a good starting point. The article suggests that "contextualized informed consent" can be useful or a method not enumerating "benign, nonspecific symptoms" with the patient's consent. This is often a practical matter given the length of rare and nonspecific symptoms included in package inserts these days. If the patient present with what I consider to be a rare side effect, I will look it up during the session, give them a ballpark estimate of how common it was in the clinical trials and discuss other potential causes.
The article also points out that side effects can also point to problems in the physician-patient relationship. Psychiatrists who are training in therapeutic neutrality often encounter difficulties with dependent patients who frequently present as not tolerating any medication or suggestion.
In addition to typical prescription scenarios this dynamic applies to other important clinical scenarios. Treating acute and chronic pain is a case in point. There is probably no better example of patient expectations based on physician advice than this prescribing relationship. If opioids are involved the scenario is complicated by a medication that reinforces its own use. The physician discussing medication response can bias the patient in the direction of dose escalation in the context of what is perceived as an incomplete response to pain when chronic pain does not respond completely to opioid analgesia.
In many cases entire illnesses are the result of problematic information supplied by physicians. Cases of chronic Lyme disease and the extremes that many of these patients go to including the application of devices that constitute medical quackery is another example. The best approach in these cases is to discuss the patient's theory of their current symptoms and how anxiety amplifies these symptoms. At that point an alternate more plausible theory for their distress can be discussed and it often results in significant improvement if not resolution of symptoms.
I can't recommend this brief article by Dr. Barsky enough. It should be read and understood by all psychiatric trainees and practicing psychiatrists. These concepts are critical to both discussions with patients about a treatment plan in order to avoid iatrogenic effects and to make sense out of somatization.
George Dawson, MD, DFAPA
1: Barsky AJ. The Iatrogenic Potential of the Physician's Words. JAMA. 2017 Dec 26;318(24):2425-2426. doi: 10.1001/jama.2017.16216. PubMed PMID: 29090307.
Figure at the top is stock photo from Shutterstock per their licensing agreement by kasezo entitled:
Stock illustration ID: 284558927 conceptual 3d design of false pill.( placebo and nocebo effect.red and green colored version)