Higher activity levels seem to be linked to lower rates of anxiety, based on a 2019 analysis of data from many scientific studies (see Schuch et al.). Emphasis on “seem.”
Their main, superficial result is that many people who exercise will still develop anxiety, but 26% fewer of them than sedentary people. The authors focused on 13 studies with “moderate to high methodological quality and a low risk of bias” with a huge total sample size of 76,000 people, and they made adjustments to eliminate the effect of gender, BMI, and smoking. That is, they tried to make sure that observed effects were actually due to differences in activity level, and not those factors.
And the answer they got was that anxiety seems to be quite a bit less common in more active people. So the simple headline “exercise helps anxiety” does seem like a fair summary. But, as usual, the details are devilish, and it’s actually not such a clear win. In fact, it’s quite a good example of how much of a difference reading the fine print can make.
With such a huge pool of data to play in, the authors decided to break it down into several different types of anxiety, and found that the results were statistically significant only for PTSD and agoraphobia (relatively rare)… and not generalized anxiety (extremely common) and a few others. Although activity seemed to help all types of anxiety, there was not actually enough data here to be sure in most cases — a data pie of 76,000 subjects seems big, but it can easily be sliced into pieces too thin to trust. It’s likely that exercise does help most types of anxiety, but it’s hard to actually know it from this data.
And this is why science is slow to be sure of much of anything that’s squishy and complicated. Did this meta-analysis actually show that exercise is not good for generalized anxiety? Or merely that they didn’t have enough of the right data to answer that more specific question? It’s not clear — despite a huge meta-analysis of a fairly well-studied subject.