Sometimes when a patient is given a treatment, like an antidepressant, the symptoms go away for awhile, and then come back. This is called a relapse. One reason for a relapse would be that the drug stops working. Another possibility, raised by this study, is that the drug was really working by means of a placebo effect. The placebo effect is something we have all heard of, but we tend to dismiss it as a problem in experiments, rather than an issue in the real world of treatments. The placebo effect is important in both settings.
In an experiment, the placebo effect is well-known potential problem, and good experiments are designed to avoid it. Here is how it works: If you give someone a new drug, or a new kind of therapy, or really any treatment at all, they usually expect it will work. After all, the person giving the treatment, be it doctor, therapist or scientist, is presumed to be some kind of expert. That expectation, that the treatment will do some good, can be a powerful force all on its own. In a depressed patient, the hopefulness that accompanies the expectation of a treatment working may be enough to improve a patient’s mood. That in turn will improve their interactions with other people. Better interactions will also help to improve their mood. So the expectation of getting better really does make the patient better. You can see why this is sometimes called one kind of expectancy effect.
The problem in an experiment is that we want to isolate one variable, the treatment effect, from any other variables, including expectancy effects. It is actually pretty simple to do. In an experiment looking at the effectiveness of an anti-depressant we need to have two groups getting pills, and one group not getting pills. One of the pill groups gets the antidepressant. The other one gets sugar pills, which are the placebos. The third group gets nothing. It is important that these three groups are otherwise as similar as possible. It is also important that those patients getting pills not know whether they are getting the real antidepressant or a placebo. We want the expectations to be the same for both groups.
Now let us imagine that the 66% of treatment group, the one that got the antidepressants. gets a lot better. That sounds good, but it is not really a measure of the effectiveness of the drug. It is really a measure of the drug effects, plus that expectancy effect. If 66% of the placebo group also got a lot better, then it looks like the drug was not effective after all, except by creating an expectation of improvement. If only 10% of the placebo group got a lot better, then it looks like the drug had a lot of benefit beyond just creating the expectation of improvement. We can use a similar logic to estimate the extent of the placebo effect, itself, by comparing the placebo group to the no-treatment group.
Now let’s switch from the world of experiments to the world of treatments. If someone who is depressed seeks treatment from a doctor, and that doctor prescribes some kind of pills, they had better not be sugar pills! The patient has a right to expect the doctor will prescribe something that has proven to be more effective than a placebo. But what does that mean about expectations? Most patients probably expect that whatever the doctor gives them will help them get better, and just like in an experiment, that expectation can be a powerful force in and of itself.
On the face of it, that should be a good thing. The patient’s expectations should combine with the drugs effects to give them a double-whammy. On the other hand, no drug, especially not an antidepressant, is going to work for everybody. And that expectation, or placebo effect, is probably not going to last forever. So what may happen is that some patients, who seemed to be doing well on whatever drug was prescribed, have a relapse some months down the line.
Here’s the question I have, and I really don’t know which answer I prefer. Should a doctor, in prescribing a new antidepressant, be as positive as possible about the potential benefit? In other words, should they try to maximize the placebo effect? Or should they strike a cautious tone, emphasizing that not every drug works for every patient?
Reference:
Zimmerman, M and Thongy, T. (2007). How Often Do SSRIs and Other New-Generation Antidepressants Lose Their Effect During Continuation Treatment? Evidence Suggesting the Rate of True Tachyphylaxis During Continuation Treatment Is Low. Journal of Clinical Psychiatry (68) 8, 1271-1276.
In order to determine the answer as to whether or not the placebo effect should be maximized, one must first inspect the nature of relief provided by an anti-depressant versus the nature of relief provided by the placebo effect. Certain clinical testing shows that the placebo effect is actually linked to the release of endorphins in patients who believe they are taking pain medications (Sauro, 2005 ). Although similar testing on depression patients has not occurred, I believe it is a safe assumption that rather than endorphins being released, a natural function against depression will occur.
The three major forms of anti-depressants are MAOIs, tricylics, and SSRIs. All three of these affect serotonin reuptake as well as a few other various neurotransmitters. (Maizels & McCarberg, 2005) Currently there is no knowledge of exactly how the decreased serotonin reuptake functions, but since there is also no knowledge of which (if any) neurotransmitters are effected through an anti-depressant placebo effect, it would be impossible to assert with any certainty that they work through similar measures. This brings about the question: Is the temporary improvement of mood by this unknown placebo effect worth the potential crash assuming the antidepressant has no actual effect, and furthermore, can simply informing a patient of the risk of failure avoid the possibility of the placebo effect?
From a personal perspective, I believe that the placebo effect should not be encouraged through means such as setting overblown standards for the patient. A patient has the right to know that failure is a possibility and must be prepared for successive treatment. Deceiving a patient would seem to be such a basic violation as to defy the Hippocratic Oath, the very core of doctor’s standards. (Farlander, 2003) By inciting the placebo effect, the doctor may also be prolonging an ineffective treatment (assuming an improvement brought about by the placebo effect would seem parallel to the expected result of the prescribed medication).
Unfortunately I also believe that the optimistic nature of many patients could lead to an unintended placebo effect, although I doubt in this specific case — depression, the prospect of an optimistic patient would arise often.
So failing the patient’s optimism surpassing the doctor’s warning, and assuming the doctor does not give the patient unrealistic expectations, a medication can be allowed to run its own effective course and thus clear results as to its success can be quickly determined. Seemingly, this avoidance of a complicated sub-path involving the placebo effect (at least in the case of depression) would lead to a more effective and safer course of treatment.
Comment by Adam Russin — September 5, 2007 @ 2:10 pm
This blog describes an experiment in which there are three subject groups, one receiving the drug being tested, another receiving a placebo, and the third receiving nothing. Even in this environment, I feel that there are biases that could confound the results of the experiment. Failing to take these variables into account might lead the experimenters to a false conclusion about their data. We have learned that there is a part of our brain that recognizes emotions/facial expressions of others. If the experimenter handing out the drugs knew which group each individual was in, he might unconsciously make some sort of facial expression that could possibly lead a subject to expect a certain treatment. In order to remove all bias from an experiment like this one, the subjects would have to be completely randomized into the three groups. This can be done using blocking techniques. The subjects might have to be blocked for gender before being put into groups. The subjects cannot know which treatment they will be receiving, otherwise there will be a lurking variable affecting the expectations of the subjects. In addition to this, the experimenter should not know what he is giving to each subject. For example, if he hands a subject a cup with the actual medication in it, and displays a slight smile, the subject might interpret the slight smirk as the experimenter’s way of saying, “too bad for you, you just got a placebo.” In turn, the subject would think that what they received would not help them get better. Just the idea that the human brain interprets the emotions of other individuals, is a lurking variable that would be extremely difficult, maybe even impossible to control.
I feel that the idea of the placebo effect also raises some morality questions. There are some debates as to whether or not placebo controlled trials of antidepressants is ethical. In an article in the New England Journal of Medicine, disagreements about placebo controlled experiments are examined. The question raised in my own head as I read the blog was addressed. “Psychological and social harms caused by depression — such as mental anguish, loss of employment, and disruption of relationships — are either not considered or dismissed.” (Emanuel, 2001) So, do the subjects in these experiments know that if they are receiving the placebo, they might experience adverse effects?
Source:
Emanuel, Miller (2001). The Ethics of Placebo-Controlled Trials–A Middle Ground. The New England Journal of Medicine. Volume 345 (12). 915-920
Comment by Brittany Sider — October 5, 2007 @ 1:51 pm
Brittany is right, of course. My hypothetical experiment would be improved if it were a double-blind randomized blocks design, as she suggests. The double blinding is especially important for exactly the reasons she points out.
And if this were a real experiment, there would be an ethical obligation (enforced by an institutional review board and ultimately by NIH) to give the patients full disclosure as part of informed consent.
Comment by intro2psych — October 6, 2007 @ 1:42 pm
In a 1996 study, psychologist Guy Sapirstein analyzed 39 studies consisting of over three thousand depressed patients. This experiment led Dr. Sapirstein to show that in the patients’ responses to the antidepressants, only 27% of the improvement was due to the medication alone. He determined that 50% was due to the placebo effect, and 23% was due to other nonspecific factors.
Using this meta-analysis as an example, it is clear that the benefits of the placebo effect vastly outweigh the possibility of a relapse. The doctor has the duty to do that which is in the best interest of the patient. By expressing even slight optimism about the drug, the doctor is able to create a placebo effect in the patient. Looking at patients that do respond positively to the drug itself, if the doctor had told them that it may not have worked, then they would not have had a strong placebo effect, and the drug would not have worked as well as it could have. Looking at patients that do not respond to the drug itself, the placebo effect increases the chances of their recovery. As long as the patient continues therapy while taking the medication, the doctor will be able to help if a relapse occurs. The power of hope in a despairing patient can sometimes make all the difference. Therefore, no matter how the patient would respond to medication, a placebo effect is still beneficial to instill in a patient, and should be maximized by the doctor.
Comment by Kevin Guckin — October 10, 2007 @ 8:15 pm