28 October 2008

What exactly is a placebo, anyway?

An alert reader recently sent me a link to this article in the New York Times, which reports that 60% of doctors have used or would use a placebo. Accompanying this report is the obligatory and entirely unsurprising whine from a bioethicist (I believe whining to be their native tongue) decrying the use of placebos as a violation of patient autonomy and informed consent.

To an extent, they are right. Patients should be informed regarding their diagnosis, and should be informed of the nature of their treatments. I don't think anybody would argue with that. There are circumstances, however, in which full disclosure is not therapeutic and may even be detrimental to the necessary alliance between physician and patient.

For the purpose of this discussion, I'd like to expand the concept of a "placebo" The denotation of the word is a medicine that does nothing. In this discussion, I'm not talking about the classic sugar pill or vitamins or other inert substances. Those are not widely used and I would say that an intervention with no conceivable medical benefit is indeed beyond the pale. In the current context, though, what I am really talking about is a real medicine which does something, perhaps something other than, in addition to, or less than what the patient thinks it does. It may be that a med has multiple mechanisms of action, or it may be that it is a real medicine which the patient did not truly need. Or there may be equipoise -- a med may be more of a nostrum that has never been proven to be effective, but has widespread acceptance as "doing something."

The central point here is to explore the limits of consent, and the degree to which it is acceptable for a physician to skirt or evade explicit consent for an intervention. Bear in mind that in most cases in the ER, there is a blanket consent obtained at registration, that things happen quickly even for non-critical patients, and it is common for interventions to be initiated before a working diagnosis is arrived at. We do not generally stop and have a prolonged discussion at each step of the process. Usually we do tell patients what we are doing, what meds we are giving, with an assumed consent. My practice is to have a fuller conversation at the conclusion of the encounter to summarize the diagnosis, the treatments in the ED, and the aftercare expectations. This assumes an uncomplicated and noncontroversial ED course, and I do modify it as needed for each individual patient.

One thing that I would like to exclude from the discussion. The linked article referred to doctors giving what I call "Go Away pills" -- antibiotics given for a viral illness, for example. I do not consider bad medicine to be a placebo. It is harmful, and not to be condoned.

Some real world examples may be illustrative of the dilemmas we encounter, and how placebos come into play in the ED:

The "worried well" patient. This patient is commonly triaged as "flu-like symptoms," in the patient's words. No fevers documented, of course. Just vague constitutional complaints, such as body aches and fatigue, perhaps an episode or two of emesis. Ultimately, there is no clear diagnosis. Now, in most cases, I am honest with these patients: I don't have a diagnosis, but I am reassured at the absence of serious signs and we can observe them without doing anything else. That goes over OK in many cases. But patients react poorly to a perceived "there's nothing wrong with you," they feel dismissed. Nonspecific diagnosis such as virus or dehydration may be well received; I may hold them out as possibilities. But often, if the ER is busy, the patent will have received a lab work-up and an IV from triage standing orders. In that case, I often will give the patient interventions such as IV fluids and/or an anti-nausea injection. This is universally perceived as therapeutic by patients. They feel very validated in their "illness" and at discharge I point to the empty IV bag as evidence that they are going to feel better now. Or maybe they are a little sleepy from the side effects of the anti-emetic and they interpret this as the medicine having been "really powerful." Now, understand that I knew they weren't dehydrated, absent reports of fluid loss, and even if they were, they could have been rehydrated orally. I probably would not have ordered an IV placed, but it was already there. I can rationalize these interventions as being potentially therapeutic, they are certainly harmless, and I know the patient will feel better. So it's totally a placebo, but is it wrong? Do I need to tell the patient in advance, "You are not dehydrated, but I think IV fluids will help you feel better"? How much honesty is necessary?

The Nostrums. Patients with chest pain can get a variety of medicines while we are doing the work-up to exclude serious causes of chest pain. One common treatment is the "GI cocktail," which is usually a mixture of maalox and viscous lidocaine. If the cause of the pain is acid reflux in the esophagus, it can be very effective in treating the pain. But it also works for pain from anxiety, pleurisy, and ischemic heart pain (beware using it as a diagnostic tool!). It tastes so nasty and creates such a strange sensation in the chest that a lot of folks are convinced that it had a potent effect and just "feel better" after its use. Sometimes we add donnatol or belladonna alkaloids, which can relax smooth muscle spasm -- that's quite useful for abdominal pain as well. Since these are effective as symptomatic relief, it's not clear that they should be described as a placebo. But since they really don't do anything to cure the underlying disease, and their mechanism of action is unrelated to the nature of the illness in some cases, it sure feels like a placebo to me. It's not supported by any strict evidence-based standards, but experience shows it to be helpful. I defend this intervention as being genuinely unclear and so I would not subject it to strict disclosure requirements.

Those are the easy examples. It gets a lot dicier when you are dealing with psychiatric complaints, or cases in which the patient's perception of the problems differs from the physician's.

An anxiety attack manifesting as chest pain. Understand that an experienced ER doc can spot this a mile away. We do work it up to exclude serious illness, but also need to treat the true diagnosis How about some ativan, a valium-like sedative? It will predictably make the patient feel better. Is it necessary to tell the patient up front that they're "just anxious," before treating their anxiety? Does the patient need to explicitly consent to an anxiolytic? How much disclosure is needed prior to treating the patient? An honest but excessively blunt diagnosis is likely to backfire, producing a hostile patient interaction and a refusal of care. Sometimes, with very careful diplomacy, in a patient with good insight and an open mind, I am able to broach this and get an explicit consent. That's optimal. But most of these folks are convinced that they are dying, are unable to accept in their native state that their illness is psychogenic, and if you even hint at that up front it's game over. In this case I often have to be less than forthright initially. I put on my grave face, validate them and make them understand that I share their concern (even though I do not); I tell them that we will be doing tests to rule out life threatening illness (which we would do anyway), but I also need them to relax and so I am going to give you something to take the edge off their nerves. It's very paternalistic, and usually accepted gratefully. Only later, once they are treated, can I tell them that they were "just anxious." Even then, diplomacy is required. But it is also clear that at least initially, there is an abrogation of fully informed consent.

Another common clinical complaint that can be positively impacted with a placebo is a condition called cyclic vomiting syndrome. This is a poorly understood phenomenon. There will be a patient who comes into the ED with intractable retching maybe ten times in a month. Then we don't see him or her for a year, and then they come in again ten times. Sometimes they are really dehydrated, but as often, in my experience, they are euvolemic. Often they are never seen to actually bring anything up, and there is a strong affective overlay to their presentations. Standard antiemetics seem not to work, but patients will often tell me that "only" dilaudid works well for them. This makes no sense. Dilaudid is a pain reliever that is well known to induce vomiting, so it's entirely paradoxical to treat vomiting with it. Many of these patients are perceived as drug seekers, and some may well be. I have a hard time figuring out which are abusers, since they all seem to present in a highly dramatic and wretched state. Due to the typical co-existent complaint of pain, opiates almost always are required to relieve the symptoms, but I have found that ativan works as well or better. Is it because this is a psychogenically driven illness? I don't know. Ativan is an effective anti-emetic; just ask any oncologist. So how much do I have to tell the patient before giving them ativan for this illness. Most patients are, quite frankly, not interested in the names of the meds. I can tell them "I'm going to give you something to stop the vomiting," and they are happy to accept. If it worked, they may ask the names of the meds later. Others will ask questions, and I give answers which are general to begin with and become more detailed as the patient's questions become more specific. Again, it's not clear to me whether this is a placebo, but my perception of why this medicine works is somewhat different from the patient's. I pretty much never tell the patient the real reason I think they were vomiting, partially because it's speculation on my part, but also because I know that it would not be well received. Patients with CVS are often "veteran" patients with fixed ideas about their illness, and there's no value for me in picking a fight that is not going to help the patient.

Let us not forget the placebo value of the therapeutic x-ray. Radiation does not help patients, but a chest x-ray or ankle x-ray is so valuable at making patients happy and feel better (relief of anxiety yet again) that thousands of doctors ignore bullet-proof evidence to routinely order x-rays on patients who do not need them. Should we inform patients that the x-ray is not necessary but if they want it they can have it?

And the most powerful placebo I have found is the deep confident voice of a tall male physician with gray hair at the temples. I have gotten good at playing that guy. I review their symptoms, discuss a differential diagnosis and why the bad things are excluded, and why I am reassured that they will be well. It's theater, pure and simple. Of course it is good medicine to be a good communicator. But it has never failed to amaze me how often a patient will hear my explanation (even when I don't do any tests), let out a deep breath, and say "Oh, what a relief. I was just terrified I had 'x'. I feel so much better now." (This is also why it's often helpful to ask in advance what the patient's real concern may be.)

In the end, I don't really know what a placebo is. There's a no bright line that separates the "sham" treatments from the "real" ones, and consent is a variable which is inconstant in terms of patients' ability or need to explicitly understand and agree to the treatments.

As an amusing side note the AP version of the article notes that the referenced study was funded in part by the National Center for Complementary and Alternative Medicine. The irony just makes my head spin.


  1. That particular article seemed somewhat misleading to me. After all, one of the questions involved giving a dextrose pill if it was shown to be more effective than no treatment; other questions involved what you called the "go away" meds. Few doctors are actually prescribing pure sugar pills.

    I'm not a tall male doctor with gray hair, but rather a short young female doctor. However, I fancy myself to be pretty good at reassuring the worried and anxious. It's all in the confidence you project, the reassurance that you've ruled out serious conditions, that their heart and lungs sound good, etc, and to have empathy with their worry. I agree that this can be rather paternalistic, but I also agree that many patients seem to be okay with this.

    And, of course, never ever use the phrase "it's all in your head".

  2. In medicine, the word placebo has a definite meaning. I don't really think your "expanded" definition has anything to do with a discussion of placebos.

    A good example of a placebo are some of these crayzee suburban hospitals that have aromatherapists on call for lavender therapy. Thing is, placebos work for a lot of this ruminating/anxiety crap and don't have any side effects. Certainly, the aromatherapist has to present this treatment in the light of how it will HELP the patient and blah blah, not "hey, this is a placebo to get you to think that we're doing something special just for you."

    Ativan is not in any expanded definition of "placebo".

  3. As an individual whose been to the ER for CVS, I can say that 100mg of amitriptyline works best for my symptoms.

  4. I wrote about this topic here.

    That article wasn't actually about placebos at all, if you read it. It was a survey that specifically avoided the word "placebo" -- and then has ethicists sounding out about how terribly dishonest their use is.

    I have a new name for CAM: Placebo medicine.

  5. when i was less experienced i would sometimes tell patients that they had a higher likelihood of having an adverse effect from the radiation exposure of an x-ray than the likelihood of finding a fracture. not well received.

    i've definitely lowered my threshold for the "go away" x-ray

  6. If you're giving a real medicine that treats the symptoms successfully, does it really qualify as a placebo? Aren't (most) patients in the ER/ED just hoping you will make them feel better?

    So, if what you do works, even if it involves waving a magic wand and intoning some special baritone doctor words, I don't see why you need to fully disclose the mechanism of action to the patient.

    Meanwhile, you've had an opportunity to evaluate them for more serious problems.

    Sounds like a reasonable solution to me (as a patient).

  7. When I worked as a volunteer EMT in New York, I discovered that plastic adhesive bandages, especially those with cute cartoon characters, worked wonderfully as pain relievers for children.

    I once had a little boy with a broken arm as a patient. I splinted his arm and we were on our way to the hospital. He was crying and complaining of pain, nothing I said or did helped. Remembering that, when I was a kid, sometimes putting a Bandaid(TM) on a small boo boo made it feel better, I asked if he wanted me to put a Bandaid(TM)on him. He nodded. We had some fun ones just for kids, so I applied one to his arm. He immediately stopped crying and when asked, said he felt better.

    I'm all for the placebo effect.

  8. I can't tell you how many patients I have who are getting ATC 8mg on zofran, have ATC compazine, and often times a daily dose of decadron that still have nausea. 1/2mg-1mg of Ativan works like a charm for them though. Yep, you're right, our oncologists like it. Not that cancer patients are usually stressed or anything...

    There's a lot of evidence that cyclic vomiting syndrome is linked to migraine. It's kind of an interesting phenomenon.

    I tend to agree that things like Ativan aren't a placebo. It's a drug that's actually treating the problem.

  9. CVS is kind of a hodgepodge diagnosis. It describes the symptoms, but the underlying causes can be varied. Indeed migraines may be one of the causative agents, which is why a TCA or a triptan might work on occasion. Other patients have metabolic disorders that make it easy for them to become ketotic. Others might have "abdominal migraines."

    All symptoms' severity is influenced by the perception of those symptoms, and this is the part that "placebo" treatments can influence. Whether it is acupuncture, homeopathy, or a deep calming voice, what you are altering is the patients' perception of their symptoms, not the underlying cause of the symptoms.

  10. "Should we inform patients that the x-ray is not necessary but if they want it they can have it?"

    Yes, absolutely. I really, really wish my doctor would be so honest with me.

    As it is, I am never sure if he is ordering a test because he is unable to make a diagnosis without it, or if it's defensive medicine, or if he simply can't figure out why I would come in to his office when (to his mind) there is clearly nothing seriously wrong with me so he's giving me a test to make me go away without complaining.

    "But it has never failed to amaze me how often a patient will hear my explanation (even when I don't do any tests), let out a deep breath, and say "Oh, what a relief. I was just terrified I had 'x'. I feel so much better now."

    You have been practicing medicine for many, many years. Why in the world would it amaze you that a patient would take your highly-educated professional opinion at face value? As you mentioned, there are things an experienced doctor can spot a mile away. We patients know this too.


Note: Only a member of this blog may post a comment.