Nocebo-Effect Noncompliance: When Telling The Patient Enough May Be Too Much
Power of Suggestion: When Drug Labels Make You Sick1 is an article on placebo and nocebo effects,2 the sort of piece that appears regularly in the popular press,with little to surprise anyone familiar with clinical practice. It is, however, a useful stepping-off point for a consideration of the impact a clinician’s manner of imparting information about treatment to a patient has on that patient’s response to the treatment and, inevitably, that patient’s adherence to the prescribed treatment regimen.
The following excerpts contain the core issues:
Research has shown that expecting to feel ill can bring illness on in some instances, particularly when stress is involved. The technical term is the “nocebo effect,” and it’s placebo’s evil twin. “It’s not a psychiatric disorder — it’s the way the mind works,” says Arthur Barsky, director of Psychiatric Research at Brigham and Women’s Hospital in Boston.
… in one 1960s test, when hospital patients were given sugar water and told it would make them vomit, 80% of them did. Studies have also shown that patients forewarned about possible side effects are more likely to encounter them. In a study last year at the University of Turin, Italy, men taking finesteride for enlarged prostates who were informed that it could cause erectile dysfunction and decreased libido were three times as likely to experience such side effects as men who weren’t told.
… Doctors may unwittingly foster placebo or nocebo effects by how enthusiastically or warily they discuss medication. “Physician communication with patients is the closest thing to magic. It gets communicated in incredibly subtle ways—a flash in the eye, a smile, a spring in the step,” says Daniel Moerman, an emeritus professor of anthropology at the University of Michigan-Dearborn. Doctors may also subconsciously transmit an expectation of pain. In a double-blind study of 60 patients who had wisdom teeth extracted, when clinicians thought they might be administering a medication that could heighten the pain instead of lessen it, the patients reported much more pain—even though they were really receiving placebos.
Should doctors discuss all those risks with patients, or can they be a self-fulfilling prophecy? It depends on the patient and the drug. “Patients should be made aware of anything that could be dangerous, so they don’t keel over on the street,” says Flavia Golden, an internist in New York City. “But if it’s minor like a headache, I don’t mention it. It’s better to keep the channels of communication open and say, ‘Call me if you have any problems.’ “
Patient-Clinician Interaction, Treatment Recommendations, Nocebo Effect, and Adherence
For the purposes of this exercise, I ask readers to stipulate that (1) the placebo and nocebo effects operate as described in the excerpts and can cause some patients to suffer significantly negative symptoms and (2) a significant number of patients will not follow a recommended treatment regimen if they suffer those significantly negative symptoms.
Now, you’re the physician recommending medication XYZ to your patient (as long as we’re stipulating, let’s also stipulate that this is the optimal medication for this patient, who will be 5 times more likely to recover from the affliction being treated than would be the case if a placebo were taken). Like most medications, there is a reasonable chance XYZ will cause some minor side-effects in a large fraction of patients and a relatively small chance it will cause a severe side-effects in a few patients.
The question is what you tell the patient and how you tell the patient about about (1) the likelihood of a favorable outcome and (2) the side-effects, including which side-effects (all, none, some) and the likelihood of those side-effects.
Keep in mind that, because of the placebo and nocebo effects, an optimistic appraisal of the outcome delivered enthusiastically is likely to cause some patients to achieve that outcome and encourage the patient to pursue the recommended treatment. Similarly, every side-effect you list for some patients will increase the likelihood they will suffer that side-effect and, as a result, increase the risk they will terminate the treatment. Of course, you can modulate that effect by how you describe the chances the side-effect will occur or how severe it may be.
It’s important to understand that the information you offer is not benign. The choice you make to tell your 100 patients taking drug XYZ that it can cause debilitating fatigue means, taking the nocebo effect into account, that some of those patients who would not have suffered debilitating fatigue will do so only because you chose to tell them about this potential side-effect. Further, some of those affected by this side-effect may well refuse to continue drug XYZ because they may think debilitating fatigue is worse than their original ailment, i.e. some patients will drop out of their optimal treatment plan only because you chose to tell them about this potential side-effect.
Are you proud of yourself?
Of course, you could decide not to tell any of them about the risk of debilitating fatigue. That’s fine – if you don’t mind hiding potentially important information from your patients and, in the process, perhaps violating ethical canons. Heck, if there is a hot shot plaintiff’s lawyer involved, you may find a jury of 12 of your peers have decided you’ve committed malpractice by not providing informed consent.
Naturally, you believe the patient must always be a partner on the clinical team so you can’t revert to the paternalistic Dr. Welby methodology of telling each patient only what you think that particular patient needs to know. And, to be fair, even if you were into retro-medical practice, unless you’re a TV doctor with good scriptwriters you won’t have more than a hunch which patients you should tell what.
Incidentally, that bit in the excerpt, “But if it’s minor like a headache, I don’t mention it. It’s better to keep the channels of communication open and say, ‘Call me if you have any problems,’ ” works better as a sound bite than as a general practice. It turns out that lots of patients are seriously miffed when they discover from the doctor taking call for you over the weekend that the headache that kept them from going to Thanksgiving dinner was a side-effect you knew could happen but chose not to tell them.3
Notice that the decision about how to inform a patient about the side-effects of a prescribed medication resolves rather straightforwardly into a forced choice about how you will manipulate the patient. Because of the placebo-nocebo effects, the reaction of some patients (and, no, you can’t tell which ones) to data, even when its transmission is seemingly neutral, is dramatically distorted. Consequently, what the clinician chooses to tell – or chooses not to tell – a patient about the side-effects of a single medication has a necessarily exaggerated impact (at least for a large number of patients), making that task, that may take place a dozen times in a morning clinic, incredibly challenging. Discussing a complex, comprehensive treatment plan for a disorder like diabetes is exponentially more difficult.
Hey, don’t look at me – I’ve been trying to figure this out for 25 years.
The only answer I’ve come up with so far is the same declaration that opens this post:
Clinicians can choose how, not if,
they influence patient compliance
- Power of Suggestion: When Drug Labels Make You Sick by Melinda Beck. WSJ November 18, 2008.↩
- A convincing argument can be made that the terms, “placebo effect” and “nocebo effect,” as they are used in the referenced article and in this post are ambiguous and counterproductive, but this is beyond the scope of today’s effort. See Stewart-Williams, S. & Podd, J., “The Placebo Effect: Dissolving the Expectancy Versus Conditioning Debate”, Psychological Bulletin, Vol.130, No.2, (March 2004), pp.324-340.↩
- I suspect the doctor quoted would agree that this is not a one size fits all sort of tactic. Quotes for articles in the lay press are seldom complete and typically are lifted out of context.↩