On Being Double Blind

Source: Jeffrey K. Smith

Every morning and every evening I either take two fish oil capsules or I don’t take any. I don’t know which. Been doing it for five years now.

It’s not that I have the world’s most selective and boring case of amnesia ever. I know I take something every morning and every evening, I just have no idea what it is. If they are “fish pills,” then they may or may not be helping me. If they aren’t, according to research, they may still be helping me through a placebo effect.

As a quantitative research psychologist with a real love for probability, I find this all mildly disconcerting. And “have no idea what it is” is a bit too strong. I have a .5 probability of taking fish oil, and an equal probability of taking something that looks just like fish oil capsules, but isn’t.

I am a subject in a double blind multi-nation multi-year study of the effects of fish oil capsule consumption on heart disease. That sounds important. I feel somewhat better just having typed that sentence. (Since I’m old enough to have a heart condition, I type instead of key enter.)

So, after four decades of conducting research, experimental and otherwise, I find myself on the participant side of the equation. And it has been something of a revelation. Consider this: Some people take fish oil because they believe it will benefit their heart condition. Other people do not take fish oil even though they have a heart condition, either believing it won’t help, or it’s too much of a bother. Both behaviors are reasonable, as the benefits of fish oil are not known.

Here is the one thing that nobody does: nobody takes something without knowing whether it is fish oil or not. Except me! And the other participants in this study. Being in a double blind study is the one status that is totally not generalizable to real life. And yet it is the gold standard for experimental treatment research. Monitor on Psychology has an excellent summary of the research on placebos along with some interesting new developments.

I wonder which group I’m in. And for reasons I don’t fully understand, I’ve come to the conclusion that I am half in each group. Or maybe both. Maybe I am Schrodinger’s subject! This is clearly a logical impossibility, and yet, that is the conclusion I’m most comfortable with. Until I know, I’m either 50/50 or both. But in thinking about my situation, I’ve also come to the conclusion that we don’t think about as many experimental conditions as we should. If we really want to test the effects of fish oil on heart conditions, shouldn’t we consider looking at the following groups?

1. People who are taking fish oil of their own accord. This would combine the effects of fish oil (if there are any), the placebo (these people think they are taking a beneficial treatment), and what might be called a “hyper-placebo effect” (“I’m taking something specifically because I believe will work.”).
2. People who are taking fish oil as part of a study and know they are taking fish oil. This would combine fish oil effect and the placebo, but not the hyper-placebo effect.
3. People who are taking fish oil, but don’t know it. This would involve deception which brings up ethical issues, but as my heart specialist said, “Hey, it’s fish oil.” This would involve only the fish oil effect.
4. People who are taking the fish oil, but think they are not. This is different from #3, where people simply don’t know that they are not taking fish oil. In this treatment, people would be told that they are not getting the treatment, when in fact, they are. Ethical problems grow, but to me, this is the most interesting experimental group.
5. People who are taking fish oil, but don’t know whether they are or not. This involves the fish oil effect, and half of the placebo effect (my situation of thinking I’m “half” in the treatment group).

We can mirror these groups with people not taking fish oil:

1. People who are not taking fish oil of their own accord. No effects.
2. People who are not taking fish oil (as part of a study) and know they are not taking it. They are taking a placebo. This would involve the strange finding (see article above) that even when people are told they are taking a placebo, it still has something of an effect.
3. People who are not taking fish oil (not part of a study), but think they are. Here we would see the classic placebo effect. I'm not even sure how you could accomplish this, much less deal with ethics, but it is a great control group.
4. People who are not taking fish oil (as part of a study), but have been told they are. Placebo effect should be in place. Ethical problems again, bigly.
5. People who are not taking fish oil, but don’t know whether they are or not. Half of the placebo effect should be taking place. Classic double blind setting.

The two #1 cases are what might be considered an observational study. There are all sorts of alternative hypothesis problems with such studies, but they are sometimes a good way to start to look at a phenomenon.

At the end of the day, I think what we truly want to compare is the following:

People who are taking fish oil of their own volition because they think it works compared to people who think they are taking fish oil of their own volition, but in fact are not. This involves deception that is probably close to intolerable, but it is clearly the best way to find out if fish oil works independent of any placebo effect.

A final note. The study is close to being over (five years!). At the end, I will get to find out which group I’m in. What am I hoping for? Let’s say the study finds out that fish oil is effective. Do I want to have had five years of benefit from it, or am I better off being in the control group and now getting to start getting the benefit of fish oil? And if it doesn't work, then I don't really care which group I was in. So you want to know what I’m thinking? I’m thinking maybe I just start taking fish oil (if it is proven effective) and not asking which group I’m in. Really.