Stretching and mortality, correlation and causation

There is still no evidence that stretching will make you live longer, despite what you may have heard — even if you heard it from a trusted source. This idea is the result of a classic example of muddling correlation and causation.

Steven Novella, in a recent post for ScienceBasedMedicine.org, made it clear that “the data is in, exercise is good for you.” He particularly emphasized that it offers great bang-for-buck, and even small improvements in fitness are disproportionately beneficial. So far, so good, and amen to all that. But then later in the post he adds this seemingly evidence-based bonus prize:

Stretching is often neglected, but has an independent correlation with longer lives and greater functionality. Just doing a few minutes of basic stretching can be helpful.

That was probably … overstated. Ironically, I’ve never heard anyone explain the foibles of correlations better than Steve!  He really gets it, and I do too thanks to him.  In this case, I suspect he was in a hurry and just didn’t give the sub-topic enough attention.

So allow me. 😜

The word “independent” is doing a lot of heavy lifting there. The implication is that all other possible explanations for the correlation have been factored out, leaving “stretching” as the only plausible explanation standing. But life and research are messy, and that’s extremely hard to establish. Whatever science it’s based on, I predicted that it wouldn’t actually do the job.

So what’s the source? He links to a Harvard blog post about the study that leads with the strong implication that boosting flexibility is an active ingredient in longevity, and not just something that came along for the ride:

“Can increasing your joint flexibility help you live longer? … it might.”

And the post is in turn about a 2024 study it does not link to (sigh), Araújo et al:

Araújo CGS, de Souza E Silva CG, Kunutsor SK, et al. Reduced Body Flexibility Is Associated With Poor Survival in Middle-Aged Men and Women: A Prospective Cohort Study. Scand J Med Sci Sports. 2024 Aug;34(8):e14708. PubMed 39165228 ❐

They followed 3,139 middle-aged subjects over 13 years, scoring their flexibility across 20 movements in 7 joints, reporting that subjects with poor flexibility had shorter lives. But this was absolutely not designed to show that the lives were shorter because of the poor flexibility.

So how about all the potential confounders? All the other things that could explain the link? Did they control for all of those? Hell no! They barely scratched the surface. They didn’t even control for diet or physical activity — the most obvious possibility. And the sample was primarily affluent white Brazilians. So reduced flexibility could easily have been a marker of poor lifestyle habits rather than a causal factor in reduced mortality. The authors acknowledge this limitation; it’s a glaring gap, and they know it. But they also dismiss it with a poor argument, and overstate the case for causality and its clinical implications.

There was nothing “independent” about this association.

The flexible subjects were probably not blessed with longer lives because they were flexible; they were probably more flexible because they had been active and health-focused, and in turn that’s likely thanks in large part to genetic gifts and luck.

This study was indeed a textbook example of “correlation is not causation.” Sometimes a correlation is indeed a valuable hint about causation, but only in the absence of more plausible explanations for the link.

I am a Dr. Novella fan, and have been for a long time, and I feel disloyal writing about something I think he got wrong when he gets so much right. But I couldn’t let this pass! There is (still) no evidence that stretching is a pillar of health and fitness, or therefore of longevity, or therefore an inoculation from the aches and pains of aging.