Massage for back pain: an interesting scientific flip-flop
Better late than never! This is something I should have blogged about in 2015. I did the groundwork back then, but never got around to actually blogging about it. It’s still an interesting topic …
For many years the best review of the science of massage for back pain available, by Furlan et al., had a cautiously optimistic conclusion, declaring that “massage is beneficial” (see Furlan 2002 and Furlan 2008). Those papers are among of the all-time most-cited scientific papers about massage.
But then, in 2015, Furlan et al. added another dozen studies to the pool of data, and actually changed their tune: now they have “very little confidence that massage is an effective treatment for LBP” (see Furlan 2015, which is the paper this post is all about).
Furlan et al.…
… in 2008: “massage is beneficial”
… in 2015: “very little confidence that massage is an effective treatment”
And nor should they have much confidence. Although there are scraps of good news, the evidence damns massage with faint praise. The swing from optimism to pessimism is fascinating, and caused some angst in the (admittedly small) community of massage therapists who pay attention to research.
think it is more that they are circumspect than pessimistic. Speaking as a scientist, we are very careful to guard against declaring a finding if there is even a small risk of it being a false positive. We never want to say ‘we’ve found something’ and later have it turn out we were wrong when more data comes in. So, I think they are hewing to scientific norms in this regard, and I do not fault them for that; it is important to be careful in science.
In this case, they may have regretted saying “massage is beneficial” based on inadequate evidence in the past, because they did indeed have to walk their optimism back as more data came in. But was it because the results were less positive? Or just that the evidence is such junk?
Garbage in, garbage out
Furlan et al. suffers badly from the “garbage in, garbage out” problem. It was not a tidy apples-to-apples review, not by a long shot.
Sometimes “more study needed” is a bullshit cover for the fact that the research so far has simply failed to produce the good news that someone wants, but sometimes more study really is needed, and I think this is a great example.
Nothing works! Back pain treatments have a long history of being disappointing
The history of medical science is littered with the carcasses of treatments that seemed “promising” based on early trials … which tend to be small and have a high risk of bias. As other researchers tackle the same question with less bias and more subjects, the null hypothesis almost always asserts itself. The harsh reality is that not many ideas in medicine pan out. But we get all kinds of promising early results anyway because of the power of bias to influence research.
We already know from Machado et al — one of the most discouraging papers in the history of back pain research — that there is no good evidence that anything works for back pain: “The average effects of [back pain] treatments … are not much greater those of placebos.” (A great paper by the way. Depressing! But great.)
Massage has been one of the last remaining hopes. Unfortunately, this review — discouraging despite being clearly inconclusive — goes a long way to reducing massage to the same level as all the rest, just another long shot. Massage for back pain is still a defensible intervention insofar as it’s cheap (self-massage), and intrinsically satisfying regardless of treatment effects, and it might be more valuable to some types of patients (with unknown characteristics that may never be known, see Saragiotto et al).
But this is still kind of a major bummer.