Portal de Periódicos da CAPES

A Conversation with … Ted J. Kaptchuk, Expert in Placebo Effects

2021; Lippincott Williams & Wilkins; Volume: 479; Issue: 8 Linguagem: Inglês

10.1097/corr.0000000000001824

1528-1132

Seth S. Leopold,

Psychological Treatments and Assessments

Few topics stir more emotional responses among surgeons than do placebo-controlled surgical trials. Dramatic studies of this design on arthroscopic knee surgery [31, 33], shoulder surgery [34], and spine surgery [3] have forced surgeons to stare down the barrel at an idea we find deeply unsettling: that something we have recommended to patients in good faith may not work. These studies may also cause us to feel shame. Perhaps we've had real complications from a procedure that turns out to be no better than a sham intervention; certainly, we've charged patients money for the privilege. For those reasons, in my experience, these trials cause otherwise evidence-driven surgeons to look for sometimes-obscure reasons why the findings should not, must not, be trusted or believed. In response to these kinds of trials, surgeons may say: "It can't be as it seems." "We don't do that procedure that way" or, with a little sleight-of-hand, "We don't do that procedure that way anymore." "The surgeons in the study didn't operate on the right patients." "My patients are different from those who enroll in randomized trials." "This procedure works in my hands, there must be something wrong with that study, I'm just not sure what it is." (The worst excuse of them all). Surgeons—sometimes known as people—can be very well defended when something dear to us is exposed or undermined. The magnitude of placebo responses varies, but it's larger than most of us realize. Placebo surgery may result in some 50% of the patients who receive it demonstrating improvements in symptoms [11]. This effect size is perhaps less surprising when we consider that pharmacological placebo effects in many conditions with subjective endpoints often come in at around 35% [9], and committing to surgery engages a patient so much more compellingly than does a pill. The idea that about half the patients who undergo orthopaedic surgery would get better if the intervention conferred no specific biomechanical benefit should be front-of-mind as surgeons evaluate the retrospective case series that populate our journals. This is particularly true when—as is so often the case—patients in those series were highly selected, outcomes tools were administered by the treating team, and follow-up was insufficiently complete or long to detect all the relevant harms associated with our interventions. We know that missing patients often are not doing as well as are the accounted-for, and nothing worsens outcomes like follow-up duration. It's worth noting that those shortcomings (known in the business as selection bias, assessment bias, and transfer bias) do not offset; they augment one another, and consistently increase the apparent benefits of treatment. The same biases influence our perceptions in the clinic, and the result of all this, both in research and in practice, is that it is all too easy to believe that we and our operations are far more effective than we and they actually are [30]. The first study I crossed paths with written by the guest featured in this month's "A Conversation with …" did not involve comparing two treatments but rather two placebos for patients with arm pain: a convincing-looking faux-acupuncture intervention (involving a retractable, blunted needle) and cornstarch disguised as amitriptyline [25]. We can come to the placebo effect later; what struck me in this study was the so-called "nocebo" effect: About one in four patients whose skin was not punctured (but who thought it was) reported a side effect to "acupuncture," and nearly one in three who didn't take any amitriptyline (but who thought they did) reported harms from the pill. The adverse effects patients encountered precisely paralleled those disclosed during informed consent; one in five patients reported drowsiness, and nearly one in five reported dry mouth from not taking amitriptyline. Our guest this month, Professor Ted Kaptchuk of Harvard University, has made a career of studying placebo treatments. I use the term "treatments" with great intention, as high-quality research has demonstrated that placebos are, in fact, treatments in every respect. They have measurable—often large—physiologic effects. Placebos don't just improve pain; they can mitigate measurable endpoints like symptoms of asthma [39], and perhaps even influence immunomodulation in patients with transplants [26]. It's possible that genomic differences among us may even make individual patients more or less likely to respond to them [12]. Prof. Kaptchuk's work has demonstrated that clinical outcomes associated with placebo effects (as opposed to the total placebo responses in randomized clinical trials [RCTs]) are more than spontaneous remission of disease, normal symptom fluctuations, or regression to the mean [21]. He argues compellingly that placebo treatments are neither unworthy nor illegitimate, and that believing them to be the results of bias or prejudice misses the point: The goal of our profession, perhaps above all else, is to relieve patients of unnecessary suffering [24]. The compounds in a placebo tablet may be inert; he's shown that the simulation of treatment that surrounds the delivered pill is anything but. While the American Medical Association (AMA) holds that purposefully using placebos for their effect in clinical care is unethical unless the patient's consent is obtained [1], Prof. Kaptchuk often has sought to try to increase the magnitude of that very effect for therapeutic purposes [18, 22, 23]. I note in that same AMA document [1], the AMA distinguishes between using placebos and providing "interventions that lack scientific foundation," something physicians seem to do all the time without any compunction. This, if nothing else, should open our minds to Prof. Kaptchuk's ideas on the topic—including his research on the neurobiological substrate of placebo effects and the patient-clinician relationship [8, 10, 29, 41]. If Prof. Kaptchuk were to take issue with anything I've summarized here, it's likely he'd question my earlier estimate of the effect size associated with surgical placebos; I'd said it was in the order of 50%. A thoughtful meta-analysis his group performed found it in some instances to be much larger than that [17]. So: Is your stuff really as good as you think it is? Or are you one of those surgeons who thinks the rules don't apply to you [30]? Join me in conversation with Ted Kaptchuk, Professor of Medicine at Harvard Medical School and director of Harvard's Program in Placebo Studies and the Therapeutic Encounter, to find out. Seth S. Leopold MD:If you could tell orthopaedic surgeons one thing about placebos in surgery, what would that be (and why)? Prof. Ted Kaptchuk: Once, after a lecture, an orthopaedic surgeon came up to me and said: "Thank you for an interesting talk. But I went into orthopaedics to get as far away from placebos as I could. I wanted to fix people." I understand his viewpoint. Orthopaedic surgeons scrutinize the musculoskeletal system to find and repair objective, measurable, and quantifiable pathophysiology, verified and guided by tools like MRI, CT, ultrasound, nerve-conduction studies, and blood tests. Much of placebo research also targets musculoskeletal disorders, but the target is chronic conditions involving central sensitization and symptom amplification. Ultimately, placebo effects are predominantly about changes in subjective symptoms [19]. So, what exactly are placebo effects? I'd say, placebo effects are the salubrious benefits that derive from participation and immersion in the rituals, symbols, and behaviors embedded in any medical encounter as distinct from the active intervention. Placebo effects rely on an assemblage of explicit and conscious behaviors (like attention, warmth, validation, acts of kindness, appearance of competence, and definitive diagnosis) as well as embodied and implicit, but not necessarily conscious, nonverbal cues (like voice; facial, eye, and bodily expressiveness; and proximity relations). Orthopaedics and placebos operate in different worlds. Is this dichotomy really so simple?Professor Ted Kaptchuk Sometimes pathophysiology and biomechanical interventions can account for and even completely cure a problem. But in most situations, the highly imprecise self-perceptions of patients also accompany an objective biological disease. On top of every objective pathology, real patients are responding in variable ways. How much of an aliment is pathophysiology and how much of it is self-appraisal response? The assumption in orthopaedics is that pathophysiology is the root. This is probably right most of the time. But there is always the patient's response. There is strong evidence suggesting placebo and nonspecific effects of rituals and behaviors can reduce and reverse aberrant symptom amplification in pain and most other subjectively measured outcomes [19]. The objective pathophysiology is critical, but there is always the patient's subjective perception in most orthopaedic situations that can amplify symptoms both before and after procedures, and that can be modulated by placebo effects or placebo-like interventions. Dr. Leopold:Placebo-controlled surgical trials are rare; uncontrolled, retrospective case series—beset with selection, transfer, and assessment biases, all of which inflate the apparent benefits of treatment—are by far the most common article type in surgical journals. Those three kinds of bias hit even harder in everyday practice[30]. Given how these biases tilt the playing field both in our journals and in what we think we see in the clinic, how can we determine whether our operations really justify exposing patients to surgical risk? Prof. Kaptchuk: There are probably more than 100 RCTs comparing surgery or other invasive procedures to some type of simulated, sham, or placebo treatment [11, 17, 38]. Reading these always send a jolt: Can this be true? These RCTs show that responses to sham surgery can be clinically meaningful. Those responses often are as large as genuine surgical interventions. Are the outcomes detected in placebo arms really placebo effects? It's unclear. Without a no-treatment control, we cannot tell whether the outcomes in RCTs are more than regression to the mean, tincture of time, or spontaneous improvement. Nor can we tell whether or what percentage of these outcomes are due to the physical therapy that follows these procedures, or the carefully prescribed medication regimens. These placebo responses probably include placebo effects, but I know of no estimate of its magnitude in orthopaedics. Still, these RCTs do tell us that many surgical procedures (at least of the types that were suspicious enough to have been tested against sham controls) are similar to or not much better than the responses to sham treatment. The problem remains: Correcting detected "pathophysiology" was not better than the rituals nor any noninterventional treatments surrounding the treatments. These procedures may not have been necessary. Henry Beecher's 1961 JAMA publication of "Surgery as Placebos: A Quantitative Study of Bias" argued that new surgical procedures should be tested in RCTs in a manner akin to pharmaceuticals [2]. His argument was mostly ignored until the 1990s when a flurry of high-profile RCTs of commonly performed surgical practices failed to show superiority over sham controls [32, 36]. Beecher's criticism, or modified forms of it, have persisted [4, 38]. These criticisms probably have merit, but I think they need more nuance and more discussion. Do we need RCTs for hip replacements? Is this similar to asking for more evidence before we equip skydivers with parachutes? How should we identify what procedures are more likely to need RCTs? I'll take the liberty of making a suggestion from outside the field: I suspect that we need a better understanding of the underlying pathophysiology. I suspect that the closer there is a one-to-one correspondence between underlying biology and subjective complaints, the less it is likely that the real surgery will be similar to sham surgery. The diagnosis of a smashed foot or hip arthritis, and the correction of those problems, seems pretty close one-to-one. We probably don't need RCTs. It might be different in some spine surgery, where the abnormalities targeted also exist in asymptomatic patients. Maybe we need more community-wide surveillance to accurately measure "normal" and "abnormal." Dr. Leopold:I'd like to go a bit deeper on that. Normal physiology can look awfully similar to pathophysiology, and common anatomic variants (or normal age-related changes) overlap commonly with common symptoms. "Black discs" in the spines of people with low-back pain, torn menisci and localized knee pain (even pain with "mechanical symptoms") in middle-aged adults, and hooked acromions in folks who have age-related tendon changes and shoulder pain all have convinced generations of surgeons to cut when nonsurgical treatment was the way to go. These things can be so tempting. How do we get around this problem? Prof. Kaptchuk: I have no easy answer. When patients have not benefited from medical and alternative medicine treatment numerous times, they often seek surgeons asking for surgical fixes. I think all physicians need to be aware that the nervous system in one direction creates nociplastic pain (or what was commonly called central sensitization and symptom amplification) and, in the other direction, the nervous systems create placebo effects that produce unimpeachable evidence that an intervention "works." I believe RCTs to test procedures on cohorts of patients have an important role in clarifying a generalized assessment of benefits and risks. I think modesty is important. I think honesty is important. But in medicine, especially in clinical situations, on the patient level, we are faced with irreducible uncertainty in moments that still require us to take action. Healers are honored because we carry that heavy burden. Dr. Leopold:What should journals do differently to deal with your many discoveries about placebos, and, in case journals don't come around, what should readers do differently when they engage with surgical journals on these points? Prof. Kaptchuk: I see journals (including CORR) increasingly showing interest in placebo studies [7]. My team has no problem finding good venues for publication. I think placebo studies have raised the visibility of nonspecific effects and the patient-provider relationship. These studies are just beginning to point us toward evidence-based and feasible ways to help patients feel better; the treatments arising from this work can deliver clinically meaningful effects that surpass what we'd expect from spontaneous improvement [23]. These innovations generally target functional diseases and medically unexplainable symptoms as well as symptom perception associated with organic disease. Dr. Leopold:Placebo effects may have less to do with the intervention (sugar pill, simulated surgery) and more to do with the mechanism behind the effect; my sense is that the effect is driven in some measure by trust and by the patient's confidence in a therapeutic relationship. This leads to two versions of the same question: How can we use trust to create an effect that, in a sense, is premised on violating that trust (using deception); phrased more affirmatively, how can orthopaedic surgeons use placebos in practice in a way you (and perhaps the AMA[1])would consider ethical? Prof. Kaptchuk: First: Deception is unethical and needs to cease and desist. Placebo researchers should not get a waiver on deception; deception hurts serious placebo researchers, stains medicine, and does not give reliable information. There is no trouble when informed consent is obtained in double-blind RCTs or in settings like one experiment in which patients explicitly provided "authorization for deception"; placebo research needs to get "honest" [21]. I agree with your ideas on trust. After all my years in placebo research, I'm reaching a tentative conclusion that trust and honesty provide the basis for eliciting placebo effects in clinical practice. Placebo effects are not about false information, bestowing exaggerated expectations or the power of positive thinking. Seriously sick patients live with uncertainty in the context of hope. Hope is tragic-optimism; it is always linked to despair. Trust, honesty, kindness, and the appearance of competence—either from conscious and/or nonconscious cues—is foundational for eliciting placebo effects. Right now, we still have not determined the exact components of the amalgam patient-physician interactions responsible for placebo effects. Will we ever? Undoubtedly, placebo effects are distinct for different patients and physicians. The response rate is highly variable depending on different diseases. I think healers doing the arduous and honest work will call forth placebo effects automatically. Importantly, my research suggests that placebo effects can happen automatically and nonconsciously. A former postdoctoral fellow of mine, now an associate professor at Karolinska Institute, performed a series of elegant fMRI experiments showing placebo effects can be evoked in nonconscious ways [14-16]. In my clinical experiments, I've shown that we can increase and decrease placebo effects by changing the rituals surrounding intervention. But I'm not sure what is done in contrived RCTs and laboratory experiments is exactly comparable to clinical care. In clinic, we should not try too hard. Placebo effects happen automatically. Remembering the doctor is actually an ingredient of the treatment is key. Remembering that patients need to "know and be known" is plenty. Caring is enough; everything follows from this. My experiments just document and measure this in scientific ways. Dr. Leopold:I understand empathic listening and caring, but I'm not clear on how I would get a patient's consent to try a placebo in actual practice. Let's imagine I want to try a placebo for a patient who has a painful joint. I'm certain there is no immediate surgical indication. How do I introduce the idea in a way that is ethical and effective? I assume it's not "I'm going to give you this sugar pill…" Prof. Kaptchuk: Great question. First, target the right patients, and the right conditions. This approach works best in patients with subjective complaints and illnesses that have no detectable pathophysiology (or complaints with pathophysiology that doesn't fully explain the complaint). And, you're right; you can't say: "Here, take a sugar pill." Instead, start by asking whether the patient has heard about placebo effects. Since most everyone has, use this as the way in. Mention that in blinded, randomized trials, people in the placebo arm often improve, and that placebos cause relevant areas in the brain to release neurotransmitters that could affect the symptoms the patient is having. You can say that physicians once believed that deception was needed for a placebo to work, but evidence now suggests that some patients can get the benefit of placebo treatment even if they know it is a placebo. Acknowledge that it sounds crazy, but offer to try it out as a "mini-experiment" if the patient is interested, and let the patient know that (s)he might feel better quickly or gradually. On the subject of it sounding crazy: It's not [21, 23]. Open-label placebos (or honest placebos) have delivered clinically important benefits to patients in conditions as diverse as irritable bowel syndrome [20], low-back pain [5, 6, 27], episodic migraine [18], cancer-related fatigue [13, 40], allergic rhinitis [37], and menopausal hot flashes [35]. In summary: Never say anything untrue, don't try to force patients to believe that open-label placebos work, and avoid giving exaggerated expectations. Patients are not stupid. Feel free to express your disbelief, and if you have any doubts, it's probably best to share them with the patient. The bottom line is that the provider must be completely honest, and if one is, there is no ethical barrier to trying an open-label placebo for the right condition in your practice. I think using placebos in this way honors a definition of medicine that recalls a more archaic idea of healing and healers, but one that remains relevant to modern medicine. Dr. Leopold:A large part of your original academic training was in traditional Chinese medicine; what can a technology-driven, intervention-heavy specialty like orthopaedic surgery learn from traditional Chinese healers? What do they get right that we get wrong? Prof. Kaptchuk: I don't think traditional Chinese healers get things right that we get wrong. That said, I think premodern medicine can remind us of at least two things we know but sometimes forget. First, for both diagnosis and evaluation of treatment, Chinese medicine builds on signs, symptoms, and patient self-report. Chinese medicine reminds us that you can build an entire edifice of healing—which patients appreciate—around focusing on symptoms. Objective measures and high technological interventions should only add to this. Second, if I assume that Chinese medicine treatments are placebos, which I don't assume, or have large components of placebo effects, which is probably true [19], it is clear to me that patients can often feel better from clinical engagement alone. Doctors provide pathways of words, care, silence, touch, hope, diagnostic tests, and behaviors that organize and reorder the illness experience [28]. It is clear that a specialty like orthopaedics fixes things by primarily relying on concrete biomechanical interventions. Chinese medicine relies on symptoms and patient-clinician interactions. But both dimensions of treatment are part of the care we can provide. When I visit China now, I don't see all the broken bodies on the streets I often encountered when I lived in the China some 40 years ago. High technology and scientific analysis only increases our ability to help.