Episode 8: I will harm

[soundcloud_ultimate track=https://soundcloud.com/adamrodman/i-will-harm]

In Episode 8 of Bedside Rounds, we explore the mysterious world of the nocebo effect, where words can literally hurt — or kill. It’s all in the mind, right?

Our story starts with one of the most interesting case reports that I’ve ever read, published in “General Hospital Psychiatry.” A young man comes into the emergency room, saying, “Help me, I took all my pills.” He’s scared, pale, sweaty, and looks very ill. He tells the doctors in the emergency room that he had a fight with his girlfriend earlier in the day, and impulsively swallowed 29 of his antidepressant, which he had gotten the day before. Just what the pills are, the doctors have no idea. The patient had enrolled in a clinical trial that he had seen on TV; during the first month he had felt his mood improve significantly. Then the argument.

The patient was in bad shape. His blood pressure was 80/40, his respirations were rapid, his heart rate was 110, and he was very lethargic. In short, he was in shock. Blood and urine toxicology testing were negative for any known toxins. His physicians inserted an intravenous catheter, and infused two liters of fluids. His blood pressure initially peaked back up, but would drop precipitously again. In all he got six liters of IV fluids over four hours — that’s almost the entire circulating blood volume of an adult man. Finally, one of the physicians from the patient’s clinical trial arrived and determined that the patient had in fact taken 29 placebos — the inert substance used as a comparison group. The patient was tearful with relief, and within 15 minutes his blood pressure and heart rate had returned to normal.

What is going on here? This is a very dramatic example of the nocebo effect, the evil twin of the placebo effect. Quick refresher — a placebo is an inert medication or procedure that is used as a control in a medical trial. The placebo effect — Latin for, I will please — is the often dramatic impact these placebos have on patients. Antidepressants, such as what the young man in our story was supposedly taking, have response rates between 30 to 50% in clinical trials. Just kidding, that’s the sugar pill that the controls got — the actual drugs have response rates of 45 to 70%. Still, it’s not too bad to be enrolled in the placebo arm.

Nocebo, conversely, means, “I will harm,” and the term started to be used in the 80s to describe the negative effects of placebos. Researchers first started to notice something fishy in drop out rates for placebos in clinical trials. In the big statin trials, for example, anywhere from 4 to 26% of participants receiving the sugar pill would have to leave the study because of negative side effects.

There are three major components of the nocebo effect. One is the expectations of the patient, two is the expectations of the physician or healer, and the third is the expectation or belief in the relationship between the two. Expectation, you can see, plays a major part. In a systematic review of migraine treatments, different placebos had side effect profiles similar to the drugs they were mimicking. For example, patients who had tricyclic placebos were three times as likely to complain of dry mouth or fatigue as those getting an SSRI placebo. In fact, dry mouth is a common side effect of tricyclics. But the patients symptoms from the same (sugar) pill yielded symptoms based on what the patients — and their physicians — expected.

Nocebos (and placebos for that matter) are not just confined to inert treatments. They can play a part in any — and maybe every — medical intervention. In one experimental trial, 50 patients with back pain were divided into two groups and then told to do simple exercises. One group was told the exercise could increase pain; the other was told nothing.  As you can imagine by now, the group who was told their pain would be worse experienced significantly more pain and was able to do fewer exercises.

So nocebos can cause icky side effects, they can cause pain, they can even cause potentially life threatening shock, as in the first example. But can they kill?

The short answer appears to be, “Yes”. Some researchers and physicians have hypothesized that “Voodoo death” is an extreme form of the nocebo effect. Doctor and anthropologist Herbert Basedow described a shamanistic “bone pointing” that he witnessed in the tribes of Australian Aborigines in 1925:

The man who discovers that he is being “boned” in indeed a pitiable site. He stands aghast with his eyes staring at the treacherous pointer, and with his hands lifted as though to warn off the lethal medium, which he images is pouring into his body. His cheeks blanch, his eyes becomes glassy, and the expression of his face becomes horribly distorted. He attempts to shriek but usually the sound chokes in his throat and all that one might see is a froth at his mouth. His body begins to tremble and the muscles twist involuntarily. He sways backward and falls to the ground, and after a short time appears to be in a swoon, but soon after he rises as if in mortal agony, and covering his face with his hands, begins to moan. His death is only a matter of comparatively short time.

Far fetched, right? But this encounter actually has all the components of the nocebo effect. Both the “patient” and the shaman have expectations about this “treatment” that causes a real physical effect — death. Are the expectations between a shaman and his practitioner that different than that between doctor and patient?

The nocebo effect is not just psychological. Researchers using FMRIs have documented that placebo pain medications reduce activation in pain-sensitive regions of the brain. Blood levels of cortisol increase in patients who are given a sugar pill they are told will increase pain. Levels of dopamine increase in Parkinson’s patients who are given a placebo that they are told will improve their motor symptoms. Even basic immune responses, such as levels of interleukin-2, can be affected by placebos. You get the picture — it’s truly psychosomatic, a psychological phenomenon that truly affects our body even at the intracellular level.

And all this leads to an obvious ethical dilemma. Part of my obligation to my patients is to inform them of all risks of any intervention that I might take. We call this “informed consent”. But the primary dictum of medicine is “primum non nocere” — above all, do no harm. Knowing about the nocebo effect –admittedly not that much — what if my informed consent actually harms my patient? Take statins for example. Up to 26% of patients withdrew from the placebo arms of those trials. By telling my patient that muscle aches are a possible side effect, I increase my patient’s chance of getting this side effect. And if she can’t take the drug because of them, she could potentially be at a higher risk of a heart attack. I read an interesting article this year in the Annals of Internal Medicine on doing a so-called “n of one” trial for statin muscle aches– giving a patient with myalgias from statins alternatively a statin and a placebo — to see if the patient can truly tolerate the drug. Spoiler alert — symptom levels didn’t change in either the statin or the placebo phase of the trial. The authors took this as an encouraging sign, and got the majority of their patients back on their statin. Can these N of 1 trials really be used to defeat the nocebo effect?

Well, it’s a novel idea, and we don’t really know yet. There have been two major approaches that have been studied. One is to focus on tolerability — that is, emphasize how many people do well with an intervention. In one study of the flu vaccine, the group that was told what percentage did NOT have side effects had fewer side effects than the group that was told what percentage did.

The other approach is “permitted noninformation”. Patient’s are given the option to not hear about mild and transient side effects (they are, of course, briefed about serious or life threatening side effects). They can also be given a list with general categories of side effects, with patients deciding on whether or not they want to hear a given category. A systematic review showed that this approach in chronic pain patients decreased average number of side effects from about 4.6 to 1.6.

So I think that the big take away from all this is that physicians need to be very mindful of the powerful relationship that we have with our patients, and how insidious the nocebo effect can be. It’s not quite the shamanistic thrall of the aborigine example, but the words we use can and do cause harm. And as one of our main missions is to “do no harm,” we have to do better.


As for our sources, we have:

Nocebo Phenomena in Medicine by Drs Hauser, Hansen, and Enck, published in Deutch Artebl International in 2012 (apoligies for butchering German,” and the accompanying editorial by the authors in the New York Times called, “Beware the Nocebo Effect.”

The Nocebo Effect: History and Physiology,” by Herbert Benson, MD published in Preventative Medicine in 1997, and finally

Nocebo effects with antidepressant clinical drug trial placebos, by Drs. Reeves, Ladner, Hard, and Burke in General Hospital Psychiatry in 2007.

Dr. Joy, et al. N-of-1 (Single-Patient) Trials for Statin-Related Myalgia in the Annals of Internal Medicine.

And finally, as always, while I am actually a doctor and I don’t just play one on the internet, this podcast is intended to be purely for entertainment and informational purposes, and should not be construed as medical advice. If you have any medical concerns, please see your primary care provider.


Leave a comment