Towards the end of last week, I was contemplating what I would be writing about for Monday. No topic had quite floated my boat, but I hated to dip into the archive of topics I’ve written about before to update a post. After all, I like to be topical whenever possible. Then what to my wondering eyes should appear (yes, I know Christmas is still two months away) but a study in the British Medical Journal by a group lead by Jon C. Tiburt at the Department of Bioethics at the National Institutes of Health in collaboration with investigators at the Osler Institute at Harvard University and the McClean Center for Clinical Medical Ethics at the University of Chicago entitled Prescribing “placebo treatments”: results of national survey of US internists and rheumatologists.
Serendipity? Who cares? The study addresses a very important aspect of science-based medicine.
In essence, Tiburt et al undertook a survey of 1,200 practicing internists and rheumatologists in the U.S. to assess their usage of placebos and their attitudes towards their use. Their results, summarized in the abstract, are as follows:
679 physicians (57%) responded to the survey. About half of the surveyed internists and rheumatologists reported prescribing placebo treatments on a regular basis (46-58%, depending on how the question was phrased). Most physicians (399, 62%) believed the practice to be ethically permissible. Few reported using saline (18, 3%) or sugar pills (12, 2%) as placebo treatments, while large proportions reported using over the counter analgesics (267, 41%) and vitamins (243, 38%) as placebo treatments within the past year. A small but notable proportion of physicians reported using antibiotics (86, 13%) and sedatives (86, 13%) as placebo treatments during the same period. Furthermore, physicians who use placebo treatments most commonly describe them to patients as a potentially beneficial medicine or treatment not typically used for their condition (241, 68%); only rarely do they explicitly describe them as placebos (18, 5%).
And their conclusions:
Prescribing placebo treatments seems to be common and is viewed as ethically permissible among the surveyed US internists and rheumatologists. Vitamins and over the counter analgesics are the most commonly used treatments. Physicians might not be fully transparent with their patients about the use of placebos and might have mixed motivations for recommending such treatments.
The ethics of placebos
The placebo effect has been discussed before on this blog, of course. Dr. Novella gave a good succinct definition the placebo effect as, “any health effect measured after an intervention that is something other than a physiological response to a biologically active treatment. Both Dr. Novella and Dr. Lipson have pointed out that the placebo effect should more properly be called “placebo effects,” because it is more than just one effect. Indeed, the effect can be so complex that it can even seemingly be observed in animals, although, as our new blogger Dr. Ramey has pointed out, the apparent “placebo effect” in animals is far more likely due to the perception and expectation of humans than in any actual placebo or therapeutic effect.
As a clinical investigator, I have to deal with issues of the placebo effect all the time. In any clinical trial, especially for symptoms with a subjective component, it is not unusual for the placebo effect to produce an apparent improvement in symptoms in 30% of subjects or more. Indeed, when it comes to science-based medicine, this effect is one reason why less rigorous studies of “alternative medicine” therapies that are arguably nothing more than a placebo (namely homeopathy, which is, after all, nothing more than vigorously shaken water) appear to show efficacy. Indeed, such effects have confounded many a study of “alternative” medical therapies, as documented by bloggers for SBM (1, 2, 3, 4, 5, 6, 7, 8). One thing that has to be emphasized, though. Placebo effects have limits. For example, I am unaware of any study demonstrating that placebo effects prolonged survival in cancer patients or resulted in objectively measurable tumor shrinkage. In this case, arguably the use of placebos is more to blind the investigators to which experimental group the patient is in rather than to provide a true placebo effect. Also, in the case of cancer, it is rarely ethical to use a placebo, as under the Helsinki Declaration, new therapies must be tested against the current standard of care, not a placebo, and it’s rare that there isn’t a standard of care for even metastatic malignancies.
One of the problems with the use of placebo treatments in routine clinical practice outside of the context of a clinical trial is that placebos pose an ethical problem for clinicians. The reason is that their usage inherently requires deception on the part of the physician and makes informed consent by the patient virtually impossible, because if a patient knows that what he or she is receiving is a placebo it might not work.Moreover, as a recent study pointed out, the most robust component influencing the strength of the placebo effect is the patient-practitioner relationship, leading to the question of whether it is an abuse of that relationship for a physician to give a patient something that is not likely to help but either strongly suggest that it will or outright lie and say that it will. On the other hand, placebos can lead to the perception on the part of a patient that his symptoms have improved, and, for conditions where symptoms with a heavy subjective component are what bothers the patient the most, relief of those symptoms is what the patient comes to a physician for. The temptation to try placebos thus becomes especially strong for chronic conditions, where science-based medicine is often less effective in giving the patient what he wants: reliable relief of pain or discomfort. Given these complexities, it’s of interest to learn whether physicians still frequently use placebos. I say “still” because in the past physicians frequently did prescribe placebos to their patients for whom they couldn’t do much else, hoping to harness the power of suggestion and the placebo effect to have the patient think he’s getting better. The attitude was, “What’s the harm?”
Such an activity could be ethically justified in an earlier era, under a medical system in which paternalism was the model upon which most medical practice was built. Indeed, a blogger by the ‘nym revere pointed out that 60 years ago shots of vitamin B12 used to be used widely as an all-purpose placebo good for what ails you. (Of course, since placebos tend to be more powerful the more invasive they are, a shot would be expected to be better at provoking placebo responses than a pill; surgery is the most powerful placebo of all.) Patients were expected to trust their physicians to know and do what was best for them, and physicians were expected to know and do what was best. If a little deception was necessary, so be it. If what was “best” for the patient necessitated withholding a cancer diagnosis because the family didn’t want grandma to know she had advanced cancer, so be it. The problem is that such deception has no place in a model of a physician-patient relationship in which the paternalistic model has become less valid and the role of the physician seen more as collaborative and as an advisor. Society has evolved such that it is now expected that the physician will tell the patient the options and help the patient reach a decision, rather than simply tell the patient what to do. A medical student blogger, Jake Young, has described the current thinking quite well and rejects the use of placebos, while Associate Professor of Philosophy Janet Stemwedel argues that under some circumstances placebo use may be considered ethical.
Does this study help us with this decision? Not a lot.
The use of “inert” and “active” placebos
One important aspect of this survey of physicians is how the use of a placebo was defined:
…for the purposes of our research we defined a “placebo treatment” as a treatment whose benefits (in the opinion of the clinician) derive from positive patient expectations and not from the physiological mechanism of the treatment itself.
Another aspect of this study is that Tiburt et al asked about “active” versus “inert” placebos. Inert placebos were placebos not expected to have any activity, such as saline or sugar pills. “Active” placebos were defined as placebos containing an active pharmacological component either unrelated or marginally related to the condition being treated (prescribing antibiotics for viral infections, for example). Finally, although the authors go to great lengths to point out how they used “nonjudgmental” questions and never used the word “placebo” in their survey and how they asked the same question in different ways in order to try to nail down consistency of responses, I’m a bit skeptical that this mattered. Any competent internist or primary care physician–any competent physician–could probably see right through the avoidance of the word “placebo” and discern the purpose of the survey. For example, one question involved a fibromyalgia patient with debilitating pain, about which physicians were asked if they would prescribe a dextrose pill if a study had shown that the dextrose pill produced better subjective relief than no treatment. I’m sorry, but only a medical moron would fail to see through the purpose of this question.
It did not surprise me that few physicians use inert placebos anymore. At least since the time I was in medical school (late 1980s) and probably at least a decade before that, this practice has generally been frowned upon as a violation of patient autonomy and for its introduction of deception into the physician-patient relationship. However, the use of “active” placebos can be even more problematic. According to this survey, active placebos most commonly included:
…over-the-counter analgesics (41%) or vitamins (38%), and some used antibiotics (13%) or sedatives (13%) as placebos.
Using antibiotics as a placebo is profoundly unethical because it violates the precept of primum non nocere, more commonly known as “First, do no harm.” The reason is that it selects for antibiotic resistant bacteria, not only potentially harming the patient but others who might have the patient’s resistant bacteria passed on to them. Prescribing antibiotics for anything other than a documented or strongly suspected bacterial infection is almost always bad medicine. Over-the-counter analgesics can also cause harm in the form of stomach ulcers (nonsteroidal anti-inflamatory drugs) or liver damage (acetominophen). True, the risk is small at the doses recommended, but it is not zero. The same can be said to be true of sedatives. However, the issues surrounding analgesics and sedatives become more complicated because often physicians rationalize their use as also being mildly physiologically effective, thus producing both a “real” pharmacologic response and placebo responses. They might have a point.
“Just do something for me, Doc!”
Another interesting question is how much of this willingness to prescribe placebos comes from patient demands that the physician “do something.” Such demands are not trivial. Patients come to us as physicians because they have symptoms that are bothering them. We as physicians in general went into medicine to help people and to relieve suffering. It is very hard to say no to a patient. A medical blogger by the ‘nym #1 Dinosaur put it well:
I read the entire article very carefully and discovered that the investigators never asked WHY the respondents recommended the “placebo” treatments that they did. That would have been illuminating. I’ll bet my annual income (ok; big spender I ain’t) that “placebo” prescribing is a response to patient demands to “do something.” I know, I know; we’re supposed to spend whatever time it takes (never fully compensated) to explain to the patient why there is no effective pharmaceutical intervention for their condition (usually after the patient has refused non-pharmacologic modalities like exercise, diet, physical therapy, etc.) Guess what: the patient then goes next door, to one of the 46% of the 57% of the 1,200, who will suggest that they take OTC vitamins or other innocuous compounds, typically describing them not as “placebos” but as “a medicine not typically used for your condition but might benefit you.” Adding insult to injury, the patient usually considers that “placebo prescriber” (the one being unacceptably dishonest) to be a better doctor than the first; the one who follows “advice from the American Medical Association, which recommends doctors use treatments with the full knowledge of their patients,” by refusing to “prescribe placebos” as described in the journal article.
Of course, the vast majority of “alternative” medicine consists of nothing more than elaborate placebos. (And, remember, the more elaborate the placebo, the more likely a placebo effect.) In addition, most “alternative medicine” practitioners give patients exactly what they want and need: Time and an ear to vent to. It’s also not as simple as that in that in any therapeutic encounter there is always likely to be a component of placebo effect, even when the physician prescribes effective medicine.
Finally, one other aspect of this study is relevant for purposes of this blog, and that is how often vitamins and supplements were used as placebos. Abel Pharmboy over at Terra Sigillata makes an excellent point when he wishes that the investigators had delved more into the use of vitamins and supplements as placebos. The reason is that it is not clear whether this category just referred to vitamins or whether it encompassed various herbs and dietary supplements advertised for various purposes, such as Echinacea or glucosamine. What I tend to wonder about along with him is whether the use of dietary supplements for placebo purposes indicates that physicians know that very few such supplements have been shown to be more effective than an “inert” placebo for any condition. Or has the “complementary and alternative medicine” (CAM) movement had its effect, and do many physicians now believe in the efficacy of many of these supplements with little or no evidence of efficacy?
The reason I ask this question is a contrarian one. As it has been pointed out before, the likelihood of placebo effects occuring is maximized when both the patient and the practitioner believe in the efficacy of the treatment being prescribed. Arguably, for purposes of placebo prescribing, the “alternative medical” practitioner would be likely to provide a more effective placebo than the physician who does not believe in (or is at least skeptical of) a given supplement’s likelihood to benefit the patient. After all, if most CAM practitioners really and truly believe in their treatments, believe that they benefit patients, and believe that they work, by whatever mechanism. It doesn’t matter to them whether the mechanism is wildly, outrageously improbable, as it is for homeopathy. It only matters that they believe it, and they do. Compared to CAM practitioners, practitioners of science-based medicine are incompetent at prescribing placebos.
The bottom line to me is that I have a grave ethical difficulties when it comes to placebo use outside the context of a clinical trial. Unless I become a true believing CAM practitioner, the use of placebos would involve my telling patients something that I don’t believe, and, like Jake Young, that is something I simply don’t see myself ever feeling comfortable doing.
J. C Tilburt, E. J Emanuel, T. J Kaptchuk, F. A Curlin, F. G Miller (2008). Prescribing “placebo treatments”: results of national survey of US internists and rheumatologists BMJ, 337 (oct23 2) DOI: 10.1136/bmj.a1938