"Can Yoga Really Cure Insomnia — or Is It Just Another Health Hype?"

As alternative therapies go, yoga is one of my favourites. Not doing yoga — I like my exercise with more brisk movements — but it's a great way to move your body if you like that sort of thing.

Unfortunately, humanity is never happy to let things stay boring and simple. Blueberries have to be a superfood, supplements have to be a silver bullet for all disease, and yoga is, according to many, not just a decent way to stretch and raise your heart rate — it's more medicinal than most drugs.

According to recent headlines, this isn't just too good to be true thinking. Media everywhere is reporting that yoga is a top exercise to improve sleep, and may be more effective than psychological therapies or other methods of exercising. If you have insomnia, the data seems to show that yoga is the treatment for you.

Except the evidence really isn't very strong. While yoga may be helpful to relieve some of the symptoms of insomnia, there's no good reason to believe that it is any better than any other form of mild/moderate exercise.

Let's look at the science.

Exercises For Sleep

The science here is quite a complex explanation, so bear with me.

The new paper that has everyone so enthused about Eastern exercise is a systematic review and network meta-analysis published in BMJ Evidence-Based Medicine. This all sounds very impressive, and in theory it makes the evidence very sound.

A systematic review is a type of study whereby scientists comb through the entire literature on a subject and carefully identify all of the studies on that particular subject. In this case, the authors looked exercise interventions for insomnia, including everything from the aforementioned yoga to lifting weights. A meta-analysis is when the authors of a systematic review combine the numbers from a range of studies into one statistical model to give us a better idea of the overall impact of one intervention. A network meta-analysis is when they then combine all of the individual effect estimates — yoga, strength training, running, etc — and compare them all against each other.

The idea of a network meta-analysis is that if you have lots of different treatments it doesn't matter if they've all been tested against each other. Instead, you look at which treatments have been tested against which other treatments, and then find the best one overall. So if running is better than psychotherapy, but weights beats psychotherapy, you can make a reasonable statistical argument that weights is better than running.

In this network meta-analysis, the authors found that most exercises performed roughly equally in terms of benefit across a range of self-reported outcomes. Cognitive Behavioural Therapy — CBT, a form of psychological assistance — was the most helpful for self-reported sleep quality. Walking, jogging, and aerobic exercises were the best for self-reported insomnia symptoms. CBT was again the best for sleep latency, which is the time it takes to fall asleep. Yoga managed to beat the other interventions when it came to self-reported total sleep time, sleep efficiency, and how people felt after waking up.

The first thing that's important to note about these results is that the authors rated almost all of them as very uncertain. Most of the studies examining exercises as a treatment for insomnia were very low-quality, and as such the results very uncertain. All of the above results come from self-report, which is a very ineffective way to measure sleep quality — better research uses what's called a sleep study. The authors did do a sensitivity analysis on only those papers that reported sleep study results, but these were inconclusive and didn't show much of a benefit for any intervention.

So why has the media gone crazy about the benefits for yoga? Well, the authors rated most of their analyses as "low" or "very low" certainty, meaning that they have little/no confidence in the results being accurate. However, there were three "moderate" certainty estimates, one of which showed that yoga was associated with an increase in self-reported total sleep time. This was picked up by the authors in their discussion, and then made it into the press release for the study, and the rest is history

None

If you look very closely, you can see in the top left-hand corner of these forest plots the green "moderate" certainty result that has led to all of the headlines. Source: The Study.

So according to the authors, we can be moderately certain that doing yoga increases the amount of time that people with insomnia say they sleep by about 100 minutes a night.

But how certain is moderately certain? You would expect some amazing, massive randomized trials behind this sort of finding, but in fact the authors have identified just four studies of yoga for insomnia in their analysis:

This small 2012 study of 44 postmenopausal women which saw 1/3 of the participants drop out and showed extremely minor differences between yoga, other stretching, and no treatment at the end of the study.

This tiny 2022 study from India that looked at yoga compared to no treatment for a month. The study has some incredibly weird data in it that would usually make it unsuitable for inclusion in any meta-analysis.

A much larger 2022 study from the same research group as the one above in India, which also has a variety of concerning data irregularities.

This 2021 paper. This is one of the most bizarre randomized trials I've ever come across. The lead author is also the only researcher involved in the study, which took place between 2003 and 2007. The other author finished his PhD in 2019 according to an online resume, and was likely in high school when the study was being conducted. The lead author is a yoga teacher and researcher who somehow did the study in the early 00s, held on to the data from this research for 14 years, and then finally submitted it for publication with his colleague in 2021. There are also a variety of oddities about the research design and reporting that make it…hard to trust fully.

The authors of the meta-analysis have somehow decided that these four trials, at least when it comes to total sleep time, provide us with moderate certainty. Personally, as someone who runs randomized trials myself and spends quite a lot of time looking into potential misconduct in the scientific space, I would say that these four trials are very low-quality and that we cannot trust the numbers in any of them.

This is one of the reasons that I always tell people to read the underlying studies in any systematic review. If you want to know what the evidence is really based on, you have to read the actual original research, not the summation from authors of a systematic review.

Bottom Line

If you look at the review, the main message is that there isn't much evidence that any exercise is better than another. Most of the interventions did well on at least one self-reported marker of insomnia, but none of them had a serious benefit for objective measures of how well people sleep. At best, we could perhaps say that people who do yoga may think that they are sleeping better, but whether they actually do is entirely unknown at this point.

The evidence for non-pharmaceutical treatments for insomnia appears to be generally bad. The authors of the meta-analysis rated almost every single area of research as "low" or "very low" certainty, and as noted above they were quite generous with their ratings. It's quite likely that a more skeptical review of the included papers would rate everything as "very low" which is equivalent to saying that the evidence tells us little/nothing at all.

If you like yoga, do yoga. The evidence suggests that it is a perfectly reasonable way to exercise, and every form of exercise is ultimately going to be good for your health. The one thing we can be absolutely certain of is that there are vanishingly few people for whom exercise is harmful — it's almost always better to do more exercise than less.

But there is still no reason to think that yoga is somehow more beneficial than any other form of light exercise.