The Placebo Effect

by Brian Dunning
April 28, 2009

Got some chronic pain? An itchy rash? Hypertension? Depression? We've got the solution for you. It's guaranteed to have no side effects, but it's an extraordinary treatment for your symptoms. It's the placebo, a completely inert and ineffective intervention, that does nothing to your body at all, except to convince you that it does. Whether it's a pill containing no medication, an expensive looking electronic device that does nothing, an inhaler that provides the same air you're already breathing, or just a manipulative treatment that doesn't manipulate anything, the medical placebo is not only a crucial component of clinical trials, but it can also be an effective medical treatment in itself. After all, placebo is latin for "I please you."

Most people have heard of the use of placebos in medical testing. When you're testing an actual new drug or device, a control group of subjects receives an identical procedure, but with a placebo instead of the new treatment. If everything is properly controlled, and neither the administrators nor the subjects know which treatment is being used, we get a good understanding of the new drug's actual effectiveness beyond what the placebo provides. But what a lot of people don't know is that even the placebo control group often fares better than groups who receive nothing at all. The fact is that when you think you're being treated, you often do experience some level of relief.

There is a very important caveat. As powerful as the placebo effect can be, it does not have therapeutic value in actually treating a physical disease. It also does not usually produce any kind of measurable improvement. Its value is almost exclusively for the reduction of symptoms that can be self-reported by the patient. Basically, it can help the way a patient feels, but it can't treat their illness. Exceptions are cases like stress, insomnia, or nausea, where there is an actual condition, but no disease agent is involved.

Placebos are not just pills or drugs. Placebo treatments can also be the application of heat, ultrasound, massage, injection, manipulation, or just about anything else you can do to a body.

Who responds to placebos? Not everyone. In fact, any given placebo treatment is usually only effective on about 1/3 of patients; although nearly everyone will respond to at least one type of placebo. And in fact, most of us do, all the time. Many over-the-counter remedies, and even some prescription drugs, have been found to have effects only comparable to a control placebo. And yet, we perceive that they work, and so we keep going back for more. For example, a review of randomized controlled trials of over-the-counter cough syrups found that no such products work better than a placebo. And so, for all practical purposes, non-prescription cough syrups are placebos. That doesn't mean they don't work; it means they only work as well as a harmless substance that you think will help your cough. And often, that effect is sufficient.

But not all placebos are equal. Some are absolutely more effective than others. It turns out that a wide variety of factors have definite influence over how well a placebo works. If the placebo is a pill, its color, size, shape, taste, and odor all have an impact. In any type of placebo administration, the manner and friendliness of its delivery affect its value. Even when and where the placebo is given can have an effect: An impressive looking office building versus the trunk of a car in a back alley.

• Blue pills are more effective than red pills for calming or tranquilizing. Red pills are more effective than blue pills for stimulation.
  
  
• Two pills have more effect than one.
  
  
• Pills with a recognized, well-known brand name and packaging are more effective than generic pills. In one large trial published in the British Medical Journal (Branthwaite A, Cooper P.), branded aspirin was more effective than generic aspirin, which was more effective than a branded placebo, which was more effective than a generic placebo.
  
  
• Expensive treatments are more effective than inexpensive ones.
  
  
• The description of the placebo's effect is also a powerful factor. Patients who receive a strong warning from the doctor about the strength of the drug have better results than patients who receive a weaker description of the drug's effect. Both groups show better improvement than patients who receive no information about the drug's effect.
  
  
• Patients who receive placebos from someone in a white labcoat get better results than when the placebo is administered by someone not wearing a white labcoat.
  
  
• Better results are obtained from placebos when the doctor spends more time with the patient explaining things.
  
  
• The drama and invasiveness of the placebo is a significant factor in its effectiveness. For example, a shot is more effective than a pill. Electric shock is more effective than ultrasound. Acupuncture is more effective than manipulation.
  
  
• Paradoxically, placebo treatments that produce unpleasant side effects are more effective than placebos with no side effects.

How do these differences affect placebo controlled trials? If I'm testing a new drug to treat depression, and my drug is a small blue pill in a generic bottle, and the control placebo is a large shiny red pill in a brand name bottle, guess what? There's a good chance that the placebo may actually produce better results than my drug, even though my drug may actually be effective. This can really happen in some cases. The characteristics of the placebo treatment are poorly selected in this example, thus it's not a valid control procedure. Designers of well-performed clinical trials know this, and an important part of their job is to eliminate such variables that can skew the results. The characteristics of the placebo should always match the active treatment as much as possible.

Sometimes, this can include mimicking any side effects that the real drug might produce. There is also a class of placebos called "active placebos". An active placebo is intended to produce the same side effects as the new treatment being tested. An active placebo does contain some active agent, just not one that is intended to treat the patient's condition. Say you're testing a new drug that has the unfortunate side effect of making its subjects dizzy and nauseous. To properly control your trial, you'd need to administer a placebo that makes people dizzy and nauseous also.

Placebos can also work in reverse to produce this same effect. If you tell a patient that an inactive placebo is a powerful drug that will produce negative side effects, it actually can. This is called a nocebo (Latin for "I will harm"). One study of 109 clinical trials in which patients were told the placebo would have adverse effects actually did produce them in 19% of the patients. These side effects were mainly headaches, drowsiness, and weakness: Again, the usual types of conditions that placebos are able to affect.

Since belief is presumed to be the major driver of a placebo's effectiveness at treating self-reportable symptoms, there is some debate over whether the effect also exists in animals. It has been shown that many animals can be calmed by soothing human contact. Dogs and horses in particular will often experience reduced heart rates when they are petted. When animals are taken to certain alternative veterinary practitioners for treatments that we would normally ascribe to the placebo effect in humans, such as veterinary homeopathy or whispering or acupuncture, the treatments are always accompanied by soothing petting by their owners and other kind treatment. Predictably, the animal's mood will typically improve; and just as predictably, the owner will often attribute this to the treatment.

Thus, you might be inclined to argue that the placebo effect isn't needed to explain the animal's improved behavior. And in many cases, you're probably right. There are often overlooked influences present that can explain an animal's improvement, or as in the case of at least one experiment, the animal reacting negatively. In this case experimenters were studying the effect of a drug injected into rats, and found that the rats were more stressed than non-injected rats. Finally employing a placebo injection of saline solution, they determined that the stress was not caused by the drug at all, but rather by the injection procedure.

Since belief in a treatment's effects is the important factor in placebos, it seems unlikely that it could apply to animals, since they lack the understanding of what the treatment is intended to do. Another complication is that many animals can be influenced by learned behavior. For example, if you give rats a mild shock when they eat from the wrong bowl, they can learn to eat from the right bowl. Take the right bowl away, forcing them to eat from the wrong bowl, and rats will exhibit stress because they associate that bowl with the shock. Some have argued that this is an example of the placebo effect, since the rat understands that this food will cause a shock. But since the effect is equally well explained by simple learned behavior and associations, the jury is still out on whether placebos work on animals. What is known is that to effectively test a treatment on animals, you must give the control group a properly designed placebo, just as you would with humans; however this is to eliminate other causes for the treatment's potential effects (like the stress from the scary injection), and it's not intended that the animal will expect to get better the way a person might.

You can actually get real placebo pills if you want them for some reason. In some locales, a pharmacist can sell them to you upon request. In other places you may need a prescription from a doctor, so you'll have to employ your salesmanship skills. If all else fails, you can buy any homeopathic pills, since those are technically placebos. I spoke with one doctor who keeps generic looking vitamin pills on hand, to give his children whenever they insist they're sick and want Dr. Daddy to fix them.

Understanding the power of the placebo effect is crucial to understanding the value of a claimed alternative therapy. If it's well designed and well delivered, an implausible therapy with no clinical value can indeed produce a subjective improvement in the patient's symptoms. To debunk a worthless alternative therapy, it's not necessary to prove that it has no effect at all. Rather, understand that under the right conditions it can, in fact, have a sometimes significant effect; an effect which can almost certainly be fully explainable as a placebo.

By Brian Dunning
Follow @BrianDunning