Narisara’s question in response to my kapalabhati post inspired me to write about this recent study: Yoga respiratory training improves respiratory function and cardiac sympathovagal balance in elderly subjects: a randomised controlled trial. I thought it would serve both as an opportunity to explore some interesting aspects of physiology and as a primer on scientific research in general.
You don’t need to know anything about science to practice yoga. The ancient yogis didn’t. Their method was introspection. They discovered the effects of yogic practices through sustained attention to their own internal physical and mental states. Given how much variation there is between individuals, that’s probably still the best way for you to determine the effects of your own practice.
However, yoga and meditation also teach us how untrustworthy our interpretation of things can be. We see what we want to see. Patanjali describes it in the Yoga Sutras as viparyaya—misconception.
One of the fundamental ways we are primed by evolution to mis-conceive is to see causality where none exists. “I practiced headstand, and my headache went away, so headstand much have cured it.” Not necessarily. Just because two events are correlated doesn’t mean one caused the other. And even if it did, that doesn’t mean it would work for anyone else.
A lot of scientific studies simply find correlations. That’s particularly true in epidemiological research, where scientists look at large populations to discover things like “people who drink coffee have higher rates of depression.” (I’m just making that up. I don’t know if there’s any such study.) The media often report those studies with headlines like “Coffee causes depression.” Well does it? Maybe. But maybe not. We can’t tell, because it’s simply an association. Correlation doesn’t prove causality. At most, it can generate hypotheses, which then have to be tested.
From a scientific point of view, the only way to demonstrate causality is in a randomized controlled trial. Take two groups of randomized individuals, who—in theory—don’t differ in any significant way. Test each group at the beginning of the experiment, then give one group some sort of experimental treatment with the other group acting as a control. Re-test everybody at the end. Assuming that the only difference between the groups was the treatment you gave—which is a very big assumption—any change in the experimental group had to have come from the treatment. However, this process can go wrong in all sorts of ways, which is why one trial doesn’t prove much of anything. In fact, to be accurate, no amount of trials can prove anything. All a scientific trial can really do is disprove a hypothesis. There’s always the possibility that another, unidentified factor caused the results we see, and not the experimental treatment.
Hopefully, that wasn’t too basic an introduction to the scientific method, but, since I plan to periodically review yoga research here, I wanted to cover the background first. The main point is that scientific studies vary in quality. Unfortunately, much of the yoga-related scientific literature is not very compelling. Many studies are poorly designed, or else so poorly reported that it’s hard to interpret their results—which amounts to the same thing from a scientific point of view.
This study, however, was well designed and reported, so I’m hoping it can serve as an example of what to look for in scientific research. Plus, you don’t have to rely on the abstract. You can read the full text here. It’s free, unlike many studies that require you either to be a subscriber to the journal or to pay an exorbitant fee to read the full text.
Anyway, on to the study. Investigators in Brazil hypothesized that practicing “bhastrika” would improve respiratory and autonomic nervous system function in healthy senior citizens. I put “bhastrika” in quotes because, while the term comes up a lot in pranayama research, it seems every researcher uses it to mean something different—probably because different schools of yoga define it differently. In this case, they mean kapalabhati with surya bhedana: 45 kapalabhati breaths, followed by an inhalation through the right nostril, a retention with bandhas, and a long, slow exhalation. (There’s a video here.)
This was a randomized controlled trial, which is generally the highest quality study design you’ll find in yoga research. (You’ll almost never see “blind” or “double blind” randomized controlled trials, which are the scientific gold standard, because it’s pretty obvious whether you’re doing yoga or not. Some studies have had control groups do “sham” yoga as a way of blinding participants as to which group they’re in, but that’s generally unconvincing.)
The groups were small, with 15 participants in the experimental group and 14 in the control. Those are pretty typical numbers for exercise training studies, because it’s expensive and difficult to recruit larger groups.
On the other hand, if the groups are too small, a few random outliers could mask an effect that’s actually there, so researchers need enough participants to ensure what’s called statistical power. It’s a good sign in a study when the authors include an analysis of statistical power, as they do here.
The authors provide a flow chart that gives some insight into the difficulties of recruiting subjects for these kinds of studies. They began with a pool of 150 participants in a yoga program for the elderly. Only 76 people volunteered for the experiment. Of those, 46 were excluded for health reasons such as cardiovascular disease, or because they were taking medications which could affect the results. That left 30 people, or two groups of 15. (One person in the control group didn’t attend the required classes and was dropped from the experiment.)
The authors also list demographic and biochemical statistics for the two groups to show that there weren’t obvious significant differences between them. If there were big differences, say one group were significantly older or had more men or women, those could affect the results. Randomization is supposed to eliminate such differences, but it doesn’t always, so it’s a good sign when researchers check. Of course, there could be other important differences they overlooked, but nevertheless, it’s a good sign.
Everyone in the study participated in an hour-long yoga asana class twice a week for four months. The experimental group followed that up with an extra half-hour of training in bhastrika, while the control group did 30 more minutes of asana. The experimental group also practiced bhastrika twice a day on their own for 10 minutes at a time, recording the sessions in a diary to verify compliance.
Most yoga studies have focused on the effect of a comprehensive practice that often includes asana, pranayama and meditation, and even lifestyle and dietary changes. That’s useful because it tells us something about the overall benefits of a yoga practice, but it doesn’t say anything about what specific exercises do. Was it the asana or the meditation that made a difference? We don’t know.
To me, the most interesting aspect of this study is that the authors chose not to do that, instead opting to test the effects of one specific practice. Everyone, both the control and the experimental subjects, did essentially the same yoga practice. The difference was that the experimental group added bhastrika. Therefore, we can expect that any differences in the results we see between groups will—probably—be due to bhastrika.
Participants were tested at the beginning of the study and again at the end. Those tests included measures of respiratory function, autonomic nervous system balance, and quality of life. I’ll wait to describe those in more detail over the next couple of posts because I also want to explore some of the underlying physiology along the way.
Part two of this series will look at how the bhastrika practice affected participants’ respiratory function.
Santaella DF, Devesa CRS, Rojo MR, et al. Yoga respiratory training improves respiratory function and cardiac sympathovagal balance in elderly subjects:
a randomised controlled trial. BMJ Open 2011;1:e000085. doi:10.1136/ bmjopen-2011-000085