Yes, it does matter what is in a placebo (as well as how it is administered, and so on) because the placebo is an important part of the experimental protocol used in pharmaceutical research. Before we get to why this question has even been raised, and an interesting point or two about it, lets quickly cover what a placebo really is.
A placebo is a tool used for making a control in an experiment. So what is a control? A control is a subset of individual variates (which may be people, or samples of some kind, or whatever) that is given a “treatment” that is just like the treatment you are testing out but missing the critical part(s) of the experiment. The control has two closely related purposes (the importance of each will vary): 1) To provide a baseline with which to compare the results of the experiment and 2) to subject the experimental variate (the person, the sample, the whatever) to things that are not part of what you are testing but that happen anyway because you are doing the experiment.
For instance, if your experiment is on humans and it requires that the humans show up at your lab for 10 minutes to have the experiment run on them, you need to rule out that some aspect of coming to your lab for 10 minutes is not the cause of whatever effect you measure. Perhaps you will be bombarding the test subject with rays from a machine you’ve invented. Well, what if your graduate students are playing around with test versions of this machine all the time so your lab tends to be bathed in the rays. Just showing up will cause anyone to be bombarded with rays, so the specific doses and methods you are using on your test subject may not be sufficiently distinct. So if you have your “controls” (a subset of your test subjects) show up and hang around for 10 minutes and leave with no treatment, you would be catching this problem.
A placebo is a special subset of control conditions where things that you think might result in a faux version of the desired effect are invoked. In the above case you may think that there is a psychological effect, where a subject who thinks they were bombarded with rays (but was not) will give a more positive assessment of their state of being afterwords than otherwise. they will say the pain from their condition is reduced, thinking that the rays must have worked. So, bombarding the patient with fake rays would be a placebo.
The baseline, or calibration, part of the placebo is one that is often overlooked when discussing placebos, and this is probably the source of some of the misunderstanding of what the so-called “Placebo effect” is.
If you get a bunch of people who have a disease that runs its course over five days, gets steadily worse then steadily better with a peak about half way through the course of the disease, and the number of days over which individuals you are looking at have had the disease so far is randomly distributed across 1 through 5, then half of the people are going to show improvement in one day because they are at the half way point or later. Thus, if you measure the level of symptoms one day and then a day later, you will see that fifty percent of the people showed improvement, even though you did not do anything.
If you have a possible treatment and simply give it to 100 people, you may find about 50 people “getting better.” That may be impressive. If you had a pill that would make half the people who are sick improve, that would be pretty good. But what really happened is that 50 people were going to get better anyway.
Here is one useful way to think about this effect. In the example I gave, the disease follows a bell-curve shaped course (or a arc over time, or something). But a real disease may follow a more complicated course, and indeed, the nature of the severity of the illness as you measure it may also depend on what you test for. If you measure how much someone sniffles vs. how lethargic they feel vs. elevated body temperature, you may get very different patterns.
What you really want to know when testing a drug is this: What is the baseline of the measured effect? With zero knowledge, one might say that the presumed baseline of effect, i.e., what will happen if your drug is utterly useless, is this: The difference between the “before treatment” symptoms and the “after treatment” symptoms is on average zero with not a lot of variation. No effectiveness = no effect.
However, the actual difference in symptoms before and after any given point in time may simply not be zero. As in the example given above, for a given disease, we may expect a large percentage of individuals to improve. Or, for that matter, get worse. Whatever.
The think is that translating the natural course of measured symptoms in a given sample of humans and predicting it in advance, in order to make a “null model” with which the sample receiving the treatment can be compared, is very difficult and probably impossible to do accurately. Instead, controls are used, and that is one of the main roles of the placebo: A treatment that is as much like the actual treatment as possible but missing, unbeknown to the recipient of the treatment or, ideally, the one giving the treatment, the ingredient you are trying to test. You give the placebo to calibrate your null model to an appropriate context-adjusted setting.
So, one part of what some call the “Placebo effect” is simply this. A second part is the idea that people who receive a treatment think the treatment must have worked and therefore report (or some how create) improvement that really isn’t there. A special version of this effect is when the belief that the treatment was given and works actually does cause an improvement, through non-mystical and scientifically verified interaction of the brain, stress (an endocrine effect involving certain relative steroid hormone levels) and the disease. This second component is often claimed, often assumed to be large, but has not been demonstrated to be very strong.
A third “placebo effect” component is the reason I felt moved to write this post. It has recently been suggested that since there is almost no standardization or regulation regarding what the contents of placebos might be in various studies, all studies that used placebos are invalid (click here if you want to be deeply annoyed). While it is true that there is surprisingly little control or agreement over what a placebo could contain, this is probably not really a big problem, and the assertion is almost certainly hooey. (See What’s in a placebo? Mike Adams certainly doesn’t know for a detailed discussion).
But there is a chance that the contents of placebos matter in a way that does not make all this research “scientificaly invalid” but that could be interesting. What if an alleged inert ingredient actually had an effect that was not known? That would actually be pretty funny. Years of research using a handful of different inert ingredients, and it turns out that one of them cures something. Unlikely.
However, consider the case of placebos in the form of pills that are taken with a glass of water. In one study that I recently read in connection with urinary tract infections, the treatment consisted of a fistful of pills taken in three or four bouts with water every day. Assume the pills were sugar pills. In this case, the test compares a treatment drug with telling someone to have, say, 8 ounces of water and two teaspoons of sugar four times a day. That could change a person’s digestive patterns. Many people probably under-drink and could use more water. In this case, adding water to the diet may have an effect! Sugar affects the immediate workings of the small intestines, which is, essentially, evolved to absorb sugar (and fat). Using oils and sugars in placebo will cause an effect … the digestive process is turned on and operated for a while. The presence of sugars in the small intestines triggers the pancreas to produce enzymes that are released at the inessential wall (where other enzymes are produced and released as well), interact with the carbs, and break everything down to simple sugars. Molecules called “glucose transporters” transport the sugars across the intestinal walls. Since these molecules will only transport sugar and salts at the same time, salt is pushed into the intestines when there is food in there. In other words, all kinds of stuff happens when you swallow a tablespoon of sugar with a glass of water.
This is probably almost always irrelevant. Recruiting the activities of the upper/middle alimentary system (mouth, stomach, small intestines) three or four times a day with a bit of extra water is not going to shrink a tumor or do much to a virus living in your liver.
However, and this is where I propose a testable hypothesis, it is possible that turning on the digestive system a few extra times a day and adding water with sugar could have an effect, of some kind, on conditions related to the digestive system. Such an effect would not (necessarily) be “curing” something, but it might be measurable. So, my hypothesis is this: Among uses of Placebos in the form of sugar pills to be taken with a certain minimum amount of water, the level of effect associated with the placebo control group will, across many studies and averaged, be larger than other cases of use of similar placebos. It would be possible to test this hypothesis with a systematic meta study. However, it would not be worth spending much research money on this for the simple reason that if this effect really does happen, the meaning of it would be more trivial than significant. But it would be interesting.