“Exercise can help people recover from depression and prevent them from becoming depressed in the first place.” NHS Choices
“Endurance exercise may help to achieve substantial improvement in the mood of selected patients with major depression in a short time.” Knubben et al (2007) Br J Sports Med 2007;41:29–33. doi: 10.1136/bjsm.2006.030130
In the last decade, the medical community as a whole has come to appreciate that regular exercise can be a real and effective way to deal with mild depression. The Mental Health Foundation did a survey of English GPs that found 56% of them thought that a programme of exercise was ‘quite effective’ in treating mild to moderate depression.
Plenty of research is being done on the topic: a quick search of the ISI Web of Science search engine (the fastest way to search for academic research papers) for ‘exercise’ and ‘depression’ reveals 22 papers written in 2012 so far that have both words in the title. These vary from neurological assessments of mice in Neuroscience to randomised trials of the effects of yoga in Complementary Therapies in Medicine. Most recently, depression and exercise have hit the headlines in the last few days following a study that suggested this link didn’t actually exist.
Exercise doesn’t help, say the headlines
“Exercise doesn’t help depression, study concludes” says the Guardian. The Daily Mail, Telegraph and Metro all carry similar stories. The study in question, published in the BMJ yesterday, is the first large scale randomised trial of its kind. It looked at the differences between patients on ‘conventional’ treatments alone (i.e. antidepressants or therapy) and those combining drugs with exercise. The results are fairly damning for proponents of physical activity.
There was no evidence that participants offered the physical activity intervention reported improvement in mood by the four month follow-up point compared with those in the usual care group … Similarly, there was no evidence that the intervention group reported a change in mood by the eight and 12 month follow-up points. Nor was there evidence that the intervention reduced antidepressant use compared with usual care.
There’s been a fair backlash in terms of anecdotal evidence from the depressed community. And I’ll admit I have not-entirely scientific reasons of my own for being sceptical of this headline. Every depressive relapse I’ve had in the past three years has been preceded by a bad cold, or a busy time at work, when I just cannot find the time to run. Regardless of what the statistical effect on the population is, running helps my depression. Period.
What does the paper actually tell us?
But I’m a scientist, so let’s do this properly. To have a reportable result it has to be statistically significant. That’s why the trial involves a large sample size. But saying that any effect is not statistically significant is not the same thing as saying that there is not an effect: only that it isn’t visible because of background noise. If I and my twin eat the same things, but I exercise and she doesn’t, and at the end of a week I’ve lost a pound and she weighs the same… not statistically significant. There are only two of us: it could be a fluke. That doesn’t mean that the exercise has had no effect: just that I can’t prove that it has. I’ve tried to draw this (rather clumsily) in the figure below. In the first pair, the difference between the means is fairly small: but so is the error, so we treat it as significant. In the second pair, the difference is much much larger. But there’s too much noise, so we treat it as not significant.
The ‘noise’ could be lots of things. Usually in scientific studies we try to either remove, or include in the analysis, “confounding variables” – or things that produce noise. This might mean separately analysing data from men and women, or only looking at participants of a particular age.
So the first thing to note is that the paper does report an effect. The difference between groups to the change in BDI score (a measure of depression) between 4 and 12 months after intervention is around 1.2 points. 1.2 points isn’t huge, but when the total difference is 3.5 points that means around a third of all the difference over the course of a year could be attributed to the exercise regime. The problem is (like the second scenario above) there’s so much noise in the system. The confidence intervals (i.e. the numbers that describe what 95% of the participants did) range from a difference of −3.42 points to +1.02 points.
Another thing I noticed was the kind of participants the study looks at. All participants had to have a BDI of 14 or more to begin with, and the control group were using antidepressants. In other words, the study wasn’t carried out on those with mild recurring depression, or those just developing depression: it looked at patients who within 4 weeks of referral already required chemical intervention or therapy. It also only included referrals from GPs, not psychiatrists, and of course self-selected for those who had not found a way to manage their depression alone.
Tuning out the noise
But most importantly in terms of the headlines the experiment only looked at two groups, and only includes one exercise regime. For instance, it is already well documented that some people experience increased endorphin or serotonin levels following exercise: but not everyone does. The study doesn’t make any attempt to differentiate between those who had enjoyed exercise prior to their depressive episode, and those who were habitually sedentary. So it’s impossible to say whether, for instance, those people who had previously exercised (suggesting that they had brain chemistry that responded positively to exercise) found exercise more helpful in treating depression than those who viewed exercise as a chore. At a more simple level, it’s impossible to say whether exercise has a different effect upon the depression of women and men, or younger and older people.
It’s also impossible to determine whether exercise might be helpful at a different level or in a different way. The study was designed to look at people doing 2.5 hours of moderate exercise a week, but actually only 15% more people in the exercise group managed that than in the control group. Really, what the study looks at is whether telling people to exercise helps them to get better faster. And the other 85% of the group were still included in the analysis. Based on the various blogs and tweets I’ve seen today and yesterday, I’m not alone in finding that the best cure for a depressive spell is a hard run: doing something less strenuous just isn’t the same. And planning to go for a run but not achieving it wouldn’t do anything at all. The authors of the paper say something along the same lines.
Perhaps an intervention that focused only on vigorous physical activity might be more effective, although we suspect that this would be less acceptable to participants and difficult or even impossible to sustain.
So what does this mean for the depressed runners?
I suppose my point (however incoherently I’ve tried to make it) is that while there is not something clearly scientifically wrong with the research to begin with, we should be wary of leaping to conclusions that “exercise is not helpful”. It would be challenging to say the least to carry out a study that included participants of both sexes, and various age ranges, carrying out a range of different types of exercise at different intensities, while maintaining sufficient participants per group to make the results statistically significant. But anything less than that does not allow a condition that exercise has no part in depression management: all we can conclude is that advising participants to undertake exercise of this type does not have a statistically significant effect upon the population as a whole. Unfortunately, I’m not sure everyone is going to think it through quite like that.