The quality of national medical reporting continues to horrify my sensibilities: there unfortunately remains a widespread lack of sophistication in the translation of medical discoveries to information that is meaningful and relevant to the general public. The most recent example is a study in the BMJ (formerly British Medical Journal) “Prescribing “placebo treatments”: results of national survey of US internists and rheumatologists” (BMJ 200;337:a1938). In this study, the authors surveyed 679 physicians (approximately 50% internal medicine physicians, 50% rheumatology specialists) with the goal of ascertaining their self-reported practices of prescribing placebos. The authors conclude that “US internists and rheumatologists commonly recommend ‘placebo treatments.’ Vitamins and over the counter analgesics are the most commonly prescribed. Physicians who use placebo treatments may not be fully transparent with their patients about their use. Where, or under what circumstances, recommending or prescribing placebo treatments is appropriate remains a topic for ethical and policy debates.”
Given the fast pace of news reporting in the age of the Internet, it would not be surprising if the vast majority of medical correspondents writing on this topic did not read the five page research article in full. The more likely case is that the writers skimmed the half-page abstract and extrapolated from the audacious claims of the authors. Accordingly, the claims expand from “half of US internists and rheumatologists commonly recommend placebo treatments” to “half of US physicians” to “US physicians regularly prescribe placebos.” While a comprehensive critical review of the research article might only be feasible to well-educated physicians and researchers with more time to mull over the details, the one paragraph “Study Limitations” section just above “Conclusions” is not out of reach of the general reader. The authors, as is required of writers publishing in respected medical journals, openly discuss the significant limitations to their study:
• “Moderate response rate“: The survey’s response rate was 57% and hardly allows for convincing generalization to the population of physicians targeted.
• The survey was attached to a broader survey on Complementary and Alternative Medicine (and the study was funded by the National Center for Complementary and Alternative Medicine (NCCAM), and Department of Bioethics, at the NIH), and the authors appropriately note that the physicians who responded to the survey may be more predisposed to prescribing placebos in association with more open attitudes toward alternative medicine practices (which often rely heavily on the placebo effect for therapeutic efficacy). Many physicians view alternative medicine practices unfavorably or with a substantial degree of skepticism, and it is less likely that this large population of physicians were represented in this study.
In addition to these limitations noted by the authors, there are other important drawbacks to the study:
• Selection Bias: The authors picked internal medicine and rheumatology, “a group of physicians who commonly treat patients with debilitating chronic clinical conditions that are notoriously difficult to manage.” In doing so, the authors are unable to convincingly generalize their findings to the wider population of physicians that includes other medicine subspecialties (cardiology, pulmonology, gastroenterology, dermatology, nephrology, endocrinology, etc.), surgical specialties, obstetricians and gynecologists, psychiatrists, neurologists, and other “generalists” such as pediatricians, geriatricians, emergency medicine physicians, and family medicine physicians. Furthermore, these patients are those that are most likely to use alternative medicine therapies due to the often insufficient treatments options provided by allopathic (Western) medicine: current medication regimens, surgical treatments, or behavioral or physical therapies may offer modest or moderate benefits without a complete (or more effective) cure (that patients are seeking). Nonetheless, the authors inappropriately generalize in their conclusions, and in making a falsely generalized claim, allow medical news reporters to sensationalize the story. The medical news reporters are equally at fault for failing to realize that these two groups of physicians do not represent the entire population of U.S. physicians. Looking at it another way, most Americans do not have “debilitating chronic clinical conditions that are notoriously difficult to manage,” and not all physicians regularly treat the patients that do. [I wonder if these medical news reporters have also written in the past about excessive subspecialization in American medicine, and yet now try to contend that internists are representative of the whole population of physicians?]
• Handpicking the Disease: The survey provided by the authors asks physicians about how they would treat fibromyalgia in a theoretical patient. The authors and medical news reporters are quick to conclude that this example is representative of the treatment of all diseases, but fibromyalgia is not. Fibromyalgia is a pain disorder of unknown etiology and, more relevant to this discussion, has insufficient treatment options at this time. A wide variety of medications are recommended or suggested as being potentially useful in the treatment of this disease, including antidepressants, pain killers (such as acetaminophen), and anxiolytics (such as benzodiazepines) in addition to behavioral modification, massage therapy, etc. The disease likely has psychiatric, neurological, orthopedic, and rheumatological aspects, but in the absence of a focused interdisciplinary effort to determine the best standard of care for this disease, internists and rheumatologists are left without standards.
Review, then, the design of this study:
Step 1: Pick a disease that is notoriously difficult to treat -> fibromyalgia
Step 2: In the design of the survey, have a theoretical situation that explains that clinical trials have shown the efficacy of a dextrose pill (a sugar pill or “placebo”, e.g. evidence-based medicine suggesting the efficacy of a drug of unknown mechanism)
Step 3: Ask the physicians if, based on this “evidence,” they would consider prescribing the medication for a disease with no universally effective standard of care
The results of the survey were moderate: 24% very likely, 34% moderately likely, 31% unlikely, 10% definitely not (remember: among a group of physicians who might not be altogether opposed to the idea of alternative medicine). Even then, the authors use this “sugar pill” example and extrapolate the results to other “placebos”: over the counter pain medications, vitamin pills, sedatives, and antibiotics. However, these medications are not necessarily placebos: they have demonstrated physiologic effects on the body that can be suspected, with reason, to have an positive effect on the management of a disease. For example, fibromyalgia is a pain disorder. Doesn’t it make sense to treat a pain disorder with acetaminophen, a widely used and available pain killer? (On the other hand, non-steroidal anti-inflammatory drugs, or NSAIDs, have not been found to be as effective and are not as likely to be prescribed.)
Notably, the authors do state they asked about specific drugs used as placebos with an explicit use of the term “placebo” in the question, and I do not contend that there are no physicians who knowingly prescribe “placebos.” However, I do not think that most physicians, even aware of the power of the placebo effect in improving patient outcomes, would regularly use placebos that might not also have a physiologic effect, such as painkillers, sedatives, or antibiotics. The authors interpret the notion that these physicians “recommend treatment primarily to enhance patient expectations” as meaning that the physicians are intentionally deceiving patients. I suspect that this issue of medical ethics is overblown: while U.S. physicians are no longer permitted by the medical establishment to withhold important treatment information from patients (as a means of influencing their decisions), it is precisely within the normal bounds of a physician’s clinical reasoning, acumen, and approach to determine how they will discuss treatment options with their patients. Is it wrong for a physician to instill positive connotations in a prescribed treatment when saying the following?:
• “This medication doesn’t work for everyone, but when it does, it has worked very well.”
• “This disease doesn’t have a lot of good treatments, but we can try one option at a time of the options we do have to see if one will work for you.”
• “It’s not known what causes the pain, but I can recommend pain medications for you to help deal with that symptom.”
Or must they revert to:
•”This treatment doesn’t work 80% of the time.”
•”This disease doesn’t have any recommended standard of care.”
•”No one knows what causes the pain in your disease. There is no evidence that painkillers help manage this symptom.” [Note: There is also no evidence that stethoscopes help improve patient care. If the study hasn't been done yet, there's no evidence!]
The authors try to address an ethical issue that they believe represents a degree of deception on the part of physicians in the patient-doctor relationship. Medical news reporters have taken the opportunity to capitalize on this story to continue a long history of attempts to degrade and criticize a profession that is more respected than their own (and generally makes more money). Neither consider the study an attempt by supporters of Complementary and Alternative Medicine to equate allopathic (Western) medicine with alternative medicine practices (relying on the placebo effect). Neither address the possibility that the underlying issue with the “placebo effect” is that its measurable therapeutic efficacy suggests that the treatment of all patients has a psychiatric component and that mental health is addressed in every treatment of every patient.
Caveat of the placebo effect: Publishing your story in a national newspaper doesn’t make it true.
[Disclaimer: I have an intellectual interest in Complementary and Alternative Medicine (CAM) practices and have previously written and argued that more appropriate research should be conducted on the efficacy of CAM therapies. I am also acutely aware of the agendas and motivations fueling research in CAM modalities as well as the experiences of CAM practitioners. I do not specifically support any CAM therapies, but I do think that physicians should be aware of CAM therapies used by their patients and that the field of medicine should find ways of considering other ideas without feeling threatened or insulted. In other words, physicians need to learn to argue intelligently and knowledgeably for or against specific CAM therapies.]