How Placebo Can Be a Self-Fulfilling Prescription
What if I told you there was a remedy that was shown to be effective for treating pain, depression, anxiety, irritable bowel syndrome, Parkinson’s disease, sexual dysfunction, schizophrenia, asthma, and migraines? It’s safer than any FDA approved drug on the market for these conditions. In fact, the more you take, the better it will work. It’s also dirt cheap–but keep in mind, the more you pay, the better it will work. It’s off-patent, so you can buy it generic, although the brand-name version works better.
Mind you, this wonder remedy is no snake oil. It’s an all-natural, vegan, non-GMO elixir with decades of clinical research supporting its use. I urge you all to walk (don’t run!) to your nearest health food store to pick up some crystalline extract of Saccharum officinarum.
Your justified reaction should be skepticism, that modern medicine could produce such a panacea. The odds that a candidate drug is shown safe and effective in humans for any one disease are meager–45% of experimental drugs fail in Phase 1 clinical trials, 60% of experimental drugs fail in Phase 2 trials, and 35% of drugs fall short in Phase 3. These aren’t mere shots in the dark–tens of millions to upwards of billions of dollars have been invested in research and development of the therapy up to that point, with often a strong biological rationale for why the treatment should work. The majority of the time, failure is due to the drug’s inability to show a benefit over placebo. Often, this isn’t a manifestation of an inactive drug, but a testament to the efficacy of placebo.
Placebo, otherwise known as crystalline extract of Saccharum officinarum(sugar pill), may be the most essential treatment in all of medicine. In a 2007 national survey, half of all doctors polled admitted to prescribing treatments they knew were ineffective for a patient’s condition, such as vitamins or over-the-counter pain relievers, in order to elicit a placebo response. These docs have long know something that clinical research has recently come to illuminate: simply expecting a treatment to work can be sufficient for the patient’s condition to improve.
Although placebos have knowingly and unknowingly been used by physicians, shamans, medicine men, and the like for most of human history, modern research into the placebo effect began in earnest during World War II. Henry Beecher, an anesthetist treating soldiers arriving from the front lines in Europe, was constantly short of supplies. As the legend goes, on one occasion Beecher had run out of morphine when a soldier needed to undergo emergency surgery. Instead of administering the potent opioid, his nurse injected the soldier with a simple saline solution. Remarkably, the soldier didn’t go into shock, nor did he gripe about the pain of the surgery. Beecher continued to administer the sham injection throughout the war while supplies of morphine were low, and found that it was 90% as effective as morphine at alleviating pain on the battlefront.
Beecher’s landmark 1955 publication The Powerful Placebo, asserted that “several classes of drugs have an important part of their action on the reaction or processing component of suffering, as opposed to their effect on the original sensation.” His was the first study to systematically quantify the response to placebo in a variety of medical contexts–including pain, headache, nausea, cough, mood disorders, and the common cold–where he estimated that treatment with placebo has a real therapeutic effect in 35% of cases.
The underlying source of the placebo response varies depending on the condition, and can be classified as either psychosomatic or interpretive. Psychosomatic effects are responses that can be objectively (although perhaps not practicably) determined to be caused by the administration of a placebo. In other words, these are true placebo effects. By contrast, interpretive effects derive from statistical artifacts or judgement errors that give the patient or caregiver the impression that the patient’s condition is improving because of placebo, when in fact the improvement is due to some other factor, or the condition is not actually improving at all. The distinction between the two is not always clear, a fact I will highlight with some mystifying examples.
The Original Sin of scientific inference is mistaking correlation for causation. A patient walks into a physician’s office complaining of aches. The physician tells the patients, “take two aspirin and call me in the morning.” Lo and behold, the next day the patients calls in to report that the heretofore crippling aches have swiftly disappeared. The patient (and a more unscrupulous physician) might be led to believe that the aspirin had acted at the site of inflammation to inhibit cyclooxygenases, causing an anti-inflammatory response that led to resolution of the patient’s pain. However, to draw this conclusion one would need to rule out several likely alternatives, including:
Regression to the mean–fluctuations in the severity of symptoms could explain why patient was suffering on Day 1 but not on Day 2.
Natural course of the disease–the illness may have resolved on its own, regardless of whether there was an intervention (think common cold).
Parallel intervention–the patient took added measures, such as getting a full night’s rest, hydrating, or eating well, that were truly responsible for the improvement.
Subsiding toxicity–the aches were an adverse effect of another medication that the patient had either stopped taking, or for which the toxicity had subsided.
Patient bias–a large category that includes the possibility that the patient was not actually relieved of aches on Day 2. For instance, the patient could be misjudging the frequency, duration, or intensity of aches in determining that there was an improvement. Or the patient may simply be lying to the physician out of deference.
This last category is where things get hairy. If a patient’s ailment has not improved by any objective measure, but the patient perceives that it has, then hasn’t it? At the very least, the patient’s experience of suffering is not as debilitating, so he or she has succeeded in managing the pain.
Confounding matters further, modern science has not yet been able to produce an objective measure of pain. In fact, Beecher was among the first to describe how the experience of pain does not correspond well to the severity of injury. He observed during the war that many gravely wounded soldiers cooly declined morphine where, from his experience, similarly afflicted civilians would have begged for it.
Of course, this is quite troubling to scientists, and a major focus of current pain research is developing a functional MRI-based biomarker for pain. In other words, if the patient can’t tell you, and the injury can’t tell you, maybe the brain can. The first places to look would be the dorsal anterior cingulate cortex and anterior insula, which contain neurons encoding for the response to harmful stimuli. However, each voxel in a typical fMRI study contains over 5 million neurons, and many neurons in these regions encode information unrelated to pain. Moreover, pain, like any sensation, is encoded in distributed circuits, so the search for pain in any one region, even at cellular resolution, may not adequately capture the underlying pain experience. The search for where pain resides may ultimately prove as elusive as our search for a locus of human consciousness.
Here, the line is blurred between the interpretive and the psychosomatic. When we can’t readily distinguish a patient’s experience of pain from his or her description of pain, then we can’t be sure whether the patient’s pain is measurably lower, or if he or she simply believes it is. And, strangely enough, it is possible that believing you’re in less pain may objectively reduce it.
Anyone who exercises is familiar with the sensation of endorphins–the body’s endogenous activators of opiate receptors that lead to mild euphoria and pain relief during physical stress. This endorphin rush derives from our hacking the body’s evolved fight-or-flight response (there’s no benefit to worrying about a splinter when you’re running from a lion). Endorphins act by the same mechanism as heroin, morphine, oxycodone, fentanyl, etc., so it’s fair to think of compulsive exercising as just a socially acceptable form of opiate addiction.
A landmark 1978 study out of UCSF implicated endorphins in the placebo-mediated relief of pain. Patients given an injection of placebo following dental surgery were asked to report the severity of their post-operative pain, and classified into either placebo responders (39%) or non-responders (61%) based on the magnitude of pain relief from placebo. The mean pain rating for responders was half that of non-responders, so the placebo effect was substantial among responders. The patients were then given a second injection, an hour later, of either placebo or the opioid antagonist naloxone(aka Narcan). Among the placebo responders, naloxone increased pain ratings to the same level as that of non-responders, while naloxone had no effect on placebo non-responders. Thus, the analgesic effect of placebo was naloxone-reversible, suggesting that endorphin activity accounts for the placebo response. This claim was further supported by the observation that injection of naloxone first, followed by placebo an hour later, reduces the likelihood of a placebo response.
At the level of our subconscious mind, our expectations of treatment and our conditioned response to the patient-caregiver relationship induce a physiological response that, at the very least, contributes to the relief of pain. And these psychosomatic effects are not limited to the sensation of pain.
Parkinson’s disease (PD) leads to the loss of dopamine-secreting brain cells, which in turn weakens the dopaminergic circuits that control, for instance, motor function. Treatments for PD (such as levodopa, dopamine receptor agonists, or monoamine oxidase inhibitors) mainly act to compensate for the loss of dopamine in the brain. When a patient is given medication, he or she has the expectation of therapeutic benefit, so it would be reasonable to expect that any drug (including placebo) would lead to the release of dopamine in the striatum–the reward center of the brain. If this dopamine can appreciably offset the deficit in the brains of PD patients, then placebo may cause an objective therapeutic response. A landmark 2001 article reported precisely this, as the authors were able to use positron emission tomography (PET) to demonstrate a placebo-induced release of therapeutic amounts of endogenous striatal dopamine in patients with PD, particularly in patients who experienced an improvement in clinical status with placebo treatment.
The brain also exerts a level of control over the immune system. Ever notice how you’re more likely to catch the sniffles while stressed or anxious? The body’s primary stress management system is the hypothalamic-pituitary-adrenal axis (HPA) axis, which maintains homeostasis under acute stress by controlling, among other things, the body’s level of cortisol, an immunosuppressant that can deter lethal overactivation of the immune system and minimize tissue damage from inflammation. A 2002 studyexploited this link between the brain and immune system to show that the placebo effect could be learned, in the Pavlovian sense. In this study, a distinctly colored and flavored drink was paired on four consecutive occasions with the immunosuppressive drug cyclosporin A, which specifically inhibits the activity of T lymphocytes. Afterward, when subjects were given the drink paired instead with a placebo, the authors observed a suppression of immune responses, such as T lymphocyte proliferation and release of interleukin-2 and interferon-γ from peripheral lymphocytes. Their bodies had been conditioned to associate the drink with the effects of cyclosporin A to the extent that the drink itself could mobilize endogenous immunosuppressive pathways. In effect, the placebo mirrored the activity of a potent drug known to be effective for rheumatoid arthritis, Crohn’s disease, psoriasis, and graft-vs-host disease.
Whether through an expectation of benefit or a conditioned response, the psychosomatic effects of placebo emerge from awareness. For instance, when an automatic infusion machine is used to deliver hidden injections of painkillers without a doctor or nurse in the room, patients needed a 50% higher dose to relieve their pain relative to when the doctor gave the drug openly at the bedside. Similarly, the anti-anxiety drug Valium has no noticeable effect on anxiety unless a person knows he or she is taking it. Painkillers don’t work as well for Alzheimer’s patients, as they are unable to formulate ideas about the future and don’t get the benefits of anticipating treatment. The awareness of treatment leads to an incremental benefit from medicine beyond the pharmacology of the drug itself–a placebo bump.
A crucial standard for separating signal from the noise in clinical studies is whether the effects of drug are dose-dependent (more pronounced with increasing dose). This standard motivated the design of a key 2008 study of placebo in irritable bowel syndrome (IBS), a chronic and often debilitating gastrointestinal disorder accompanied by pain and constipation. 262 patients were divided into three groups: a no-treatment group, a group receiving sham acupuncture with limited interaction with the practitioner, and a group receiving sham acupuncture with at least 20 minutes of “augmented” care. In the third group, practitioners were required to touch the hands or shoulders of patients, spend at least 20 seconds lost in thoughtful silence, and make remarks such as “I’m so glad to meet you,” “I know how difficult this is for you,” and “This treatment has excellent results.” It was the first study to show a dose response for placebo–the more care people received, the better the outcome on average.
In general, the more conspicuous the placebo, the more aware the patient is that he or she is being treated, and the stronger the placebo response. Larger pills seem to work better, two pills are better than one, and brand-name pills are superior to generics. Capsules are stronger than pills, and injections are more effective than either. Patients tend to have better outcomes when their doctors are more attentive, or when they are paying more for treatment. Colors are also relevant to the type of treatment–blue pills are effective sedatives, while red and orange pills work well as stimulants. Green is effective at reducing anxiety, while yellow is better for depression. Fake surgeries are even better than fake drugs–in the case of surgeries to relieve pain, there is essentially no difference in outcomes between fake and real surgery.
There is a dark side to the placebo effect: its evil twin, the nocebo effect. In any clinical trial, patients will experience a variety of adverse events, some serious enough to compel patients to drop out. Strangely enough, upon unblinding of the trial, many of these patients turn out to have been in the placebo arm. True nocebo effects (distinct from interpretive effects described above) tend to be similar among trials and include headache, nausea, fatigue, dry mouth, dizziness, and tachycardia. Many of these effects will be quite similar to the known side effects of the investigational drug, which isn’t entirely surprising given that participants are often warned of these potential side effects before enrolling. Similar to placebo effects, nocebo effects can arise from expectation or conditioning. For instance, cancer patients undergoing rounds of chemotherapy can suffer from anticipatory nausea and vomiting (ANV) conditioned by their previous experience of treatment. This distressing nocebo effect can be unlearned using a technique called “overshadowing”–pairing a distinctly flavored beverage with chemotherapy infusion for several cycles leads to reduced ANV during the cycle for which the drink is not provided.
Nocebo effects can have a treacherous impact on clinical trials, especially when trials are not placebo-controlled or not double-blinded, as doctors may attribute adverse effects to drug that are actually nocebo responses. These effects may discourage patients from undergoing treatment. They can fool patients into blaming their symptoms on the medicine they’re taking, and cause them to abandon a potentially beneficial treatment. In addition, they lead to waste in the healthcare system when patients seek relief from nocebo symptoms or ask to have the medication changed.
The placebo response is an integral component of almost any medical treatment, although it can be quite distinct from the pattern of response to the active pharmaceutical ingredient (API). For instance, the placebo response in depression tends to be abrupt, occurs early in treatment, and is less likely to persist, whereas the response to antidepressants tends to be gradual, occurs after several weeks, and is more likely to persist. Although as a society, we tend to attribute a therapeutic effect inordinately to the API, this finding highlights the importance of managing the entire context of treatment to the practice of medicine. By harnessing the potential of placebo to improve outcomes, we can design better therapies by optimizing for the placebo response as well as the response to API. Indeed, the ideal therapeutic for depression would be both fast-acting and have an enduring effect, which may be achieved by the combination of an effective placebo and an SSRI antidepressant.
Similarly, we tend to ascribe much more value to objective measures of patient benefit than to subjective outcomes. Unfortunately, in doing so we may miss the forest for the trees. Although quality-of-life and pain are more difficult to quantify than physiological biomarkers, they tend to be more meaningful to patients. These subjective measures also happen to be those for which placebo moves the needle. In a 2011 study, asthma patients were randomly assigned to treatment with either an albuterol inhaler (a drug that opens the airways), a placebo inhaler, sham acupuncture (where needles are withdrawn before they touch the skin), or no intervention. Two endpoints were assessed: maximum forced expiratory volume (FEV; an objective measure of lung function), and self-reported ratings of improvement. Based on FEV, only albuterol was effective compared to no intervention. However, based on self-reported improvement, albuterol, placebo, and sham acupuncture were all equally effective and significantly better than no intervention.
Effective medicine is both an art and a science, and the art of the placebo is in understanding how to incorporate it into treatment. The primary dilemma is one of ethics–how can we properly exploit a tool that relies, at least in part, on deception? The answer may lie in removing deceit from the formula. In a 2010 study, 80 IBS patients were randomly assigned to treatment with either a placebo pill twice a day or no intervention. Before the study began, both groups were informed that placebos were “inert or inactive pills, like sugar pills, without any medication in them.” They were also given the true statement that placebos “have been shown in rigorous clinical testing to produce significant mind-body self-healing processes.” Patients who received open-label placebo scored significantly higher on multiple standardized scales of IBS symptom improvement, reporting twice as much symptom relief as the no-treatment group. Even when patients know they are receiving placebo, the effect is comparable to what one would expect from an approved drug. Such studies provide a template for how the placebo effect may ultimately become a self-fulfilling prophecy, allowing us to pull the tablecloth from under the setting.
There are two final points to consider in appreciating the power of the placebo response. First, the majority of drugs we use exert their effects on receptors and pathways that have their own endogenous ligands. In fact, many drugs are simply formulations of the endogenous biomolecules themselves. The physiological levels of these biomolecules are often controlled by the autonomic nervous system, part and parcel of the unconscious mind–meaning the mind can, to varying degrees, access pharmacology similar to active drugs. Second, many approved pharmaceuticals address only the symptoms of an underlying disorder; symptoms are but tools of your unconscious mind to alert you to an illness. For two patients with the same underlying pathology at the same stage of progression, one’s symptoms may be manageable while the other’s are debilitating. The unconscious mind of a drug’s recipient defines a personal reality that plays a major role in determining the interaction of that drug with the recipient’s body.
There is some danger that uncritical acceptance of the placebo effect could be used to justify useless and potentially harmful snake oil remedies. To reject a proven therapy in favor of a pill that makes a patient feel better, but neglects the underlying source of illness, would be to do harm. The contribution of the unconscious mind to the observed action of a drug is significant, and in evidence-based settings it can be used as an ally to improve medical care, especially in the context of palliative care or neuropsychiatric diseases. Placebo will never shrink tumors or eliminate viruses, but if the ultimate goal of medicine is to ease suffering, then any practitioner could benefit from a healthy dose of perspective.