We've been talking about neuronal signal transmission, the monoamine hypothesis, and the theory of how SSRIs work. Much of the recent controversy about medication rests on the examination of medication outcome trials. So let's take a time out and consider the nature of these studies.
The double-blind placebo control trial has, for better or worse, become the gold standard for the evaluation of healthcare interventions – in the areas of both mental and physical health. These studies sound straightforward enough, but like most subjects in science, things get more complicated the closer we look.
The nature of placebos
A placebo is an inactive treatment - one that no theory would suggest should work from its physical effects alone. Placebos are often referred to as sugar pills, but in fact they might be made of almost anything, so long as the substance is not believed to have any effect on the disorder being examined.
Placebos may not even be pills. Placebo surgery (in which an incision is made but nothing else is done), placebo psychotherapy (in which a clinician meets with a person but doesn't do anything believed to be effective), and placebo light therapy (in which people sit before a light box that doesn't really give off enough light to be helpful) have all been used.
Usually the placebo treatment is made to seem as similar as possible to the other (hopefully active) treatment being tested. So when you're testing the effectiveness of a medication in pill form, your placebo comparison would be a pill of the same colour, size, and shape.
The Confirmatory Bias
Why not just compare a new medication with doing nothing? That, after all, is the usual alternative in healthcare: Treat the person with something thought to be effective, or watch and wait to see whether the problem resolves itself. Why bother with a placebo control?
Human beings are prone to a powerful confirmatory bias. This means that when we develop a theory about something, we automatically assign greater significance to information that conforms to our theory than to information that contradicts it.
If I develop a conviction that rain tends always to fall on my days off, I will notice the sunny days when I am working and the rainy days on the weekend. I will tend to forget the rainy days when I am working, and the few sunny days when I am not. Over time, my theory will seem more and more true.
Similarly, if I think that life goes better when I wear blue socks, I will begin to notice that this seems more and more true. If I think that Vitamin Z cures colds, I will notice my recovery when I take it, and the persisting symptoms when I don't. When we look for evidence of almost anything (unusual behaviour at the full moon, personality types associated with star signs, mood changes associated with certain crystals), we tend to find it. Disconfirming evidence tends to be ignored.
There is a lot of uncertainty in healthcare, but we know this: the expectation of improvement will lead to a perception of improvement - in at least some symptoms, in some disorders, for some period of time. In general, the symptoms most affected will be subjective ones - things like mood, pain, and discomfort.
Placebos do not work with everything. If we give placebos to patients with high cholesterol, this does not seem to reduce their cholesterol over time. Objective measures like blood tests, tissue regrowth, and tooth decay do not seem to be nearly as affected by placebos as subjective indicators. There are exceptions to this principle, but let's leave those alone for the time being.
The Rationale for Placebo Trials
When we administer a treatment for a disorder, we can say (with a bit of oversimplification) that there are two effects:
- The genuine, measurable effect produced by the expectation of receiving a treatment for the problem.
- The physical impact of the treatment on the problem, regardless of what the client expected.
So if we violate all ethics and sneak pain medication into the food of a person with back pain (producing effect 2 but not effect 1), they should notice a reduction in pain even though they don't know why it is happening. And if we manufacture sugar pills that look like powerful pain medication (resulting in effect 1 but not effect 2), they should also reduce the pain to at least some extent.
And if we show the person the pill bottle, describe the predicted effects of the medication, and provide a pill containing the active ingredient, we will get both effects: expectation plus physical impact.
The question, obviously, is to what extent the improvements we see are attributable to Effect 1 versus Effect 2.
Perhaps our supposed miracle cure is actually no better than placebo, and all of the improvement we see is the product of expectancy. Or maybe we have made a real discovery, and the improvement is partly expectancy but partly physical impact.
Why do we care whether an effect is due to expectancy?
A placebo effect, after all, is still an effect. It's not imaginary, it's real. Do we care whether it is produced by expectancy or by the chemical ingredient?
Well, yes, for a number of reasons.
- Placebo effects tend to be somewhat time-limited in their action, and may make a disease process seem to improve while the underlying pathology continues.
- Most medical treatments have disadvantages as well as advantages. A medication that alleviates cholesterol might strain the liver, for example. If the chemical ingredient of our medication does no good but causes harm, we need to know this.
- There's an ethical issue involved in charging large sums of money for inactive ingredients.
Plus, the healthcare professions pride themselves in having treatments that are more advanced than the superstitions of centuries past. If our new patented medication is no better than repeating an incantation three times, then why do we fund medical healers but not faith healers?
The double-blind trial
In a double-blind trial, neither the person taking the placebo (the patient) nor the person giving it (the clinician) knows which medication the person is receiving.
It's obvious by now why the patient shouldn't know which treatment they are getting: This controls for expectancy, or at least it should. (There are some problems with this reasoning, but I'll leave that for another post.) So people who receive Treatment A should have about the same expectation of improvement as those receiving Treatment B. If we have succeeded in keeping expectancy the same in both groups, then any differences we see should be attributable to the chemical ingredient in the pill.
But why do we need the caregiver to be blind as well?
We know that people communicate a great deal by their nonverbal behaviour. Presumably no one in a blind trial would announce to the patient, "You're in luck - you get the real drug!" But they might do so involuntarily. If they believe that the drug is effective, they might show more anxiety and less optimism when seeing people who are taking the placebo. The result could be that patients in the placebo group would develop less powerful expectancy effects than those in the active treatment group.
As well, some of the outcome measures are collected by the treating clinician. For example, one of the outcome measures for many depression trials is a physician-rated scale. If the clinician believes in the effectiveness of the medication, he or she might see greater improvement in people receiving the active treatment than in those receiving placebo - even if they are trying their hardest to be honest and even-handed. Similarly, if they believe the medication doesn't work, they may fail to see genuine effects in the drug condition.
Consequently, it's important that the clinician - and anyone else doing evaluations of patients - not know which treatment a given patient received (particularly if there is any subjective element whatsoever to the outcome measure).
Sounds great. But...
The problem is that placebo control trials are not quite as simple as advertised. A lot of assumptions are made about how they work, and it is easy to make interpretations that aren’t warranted. In upcoming posts we’ll explore several of these:
- In a double-blind trial, neither patient nor clinician knows which condition the patient is in.
- Differences between placebo and medication can be attributed to the chemical effects of the medication.
- The placebo effect is the degree of improvement seen in the group of patients given the inactive medication.
- Expectancy is the only determinant of improvement in the placebo group.
In fact, none of these statements is necessarily true. Stay tuned.