For chronic back pain, exercise dose matters more than exercise type

Exercise is one of the most commonly recommended treatments for chronic low back pain, yet the trials behind it are wildly varied. Programmes differ in duration, frequency and dosage, in whether anyone supervises the sessions, and in whether a home-based element is included. Pooling such different trials uncritically can produce misleading conclusions about whether exercise works at all. This analysis set out to measure the effect of exercise on pain and disability in chronic low back pain, and, more unusually, to explain why the trials disagree with each other so much.

The researchers searched six databases up to August 2008 for randomised clinical trials of exercise for nonspecific chronic low back pain, where pain and disability were measured on continuous scales. From every included trial they extracted baseline demographic data, the features of the exercise programme and the outcome data. They then ran meta-regressions testing whether 8 study-level variables were associated with the size of the benefit, including baseline symptom severity, the number of exercise hours and sessions, supervision, individual tailoring, a cognitive-behavioural component and markers of trial quality.

When all types of exercise were analysed together, exercise produced small but significant reductions in both pain and disability compared with minimal care or no treatment. The more surprising result was what failed to matter. Despite the many plausible sources of disagreement between trials, features such as supervision, individual tailoring and a cognitive-behavioural component were not significantly associated with the size of the effect. Only dosage, the sheer amount of exercise delivered, was significantly linked to how large the benefit was.

The headline finding, a small but real benefit of exercise for chronic low back pain, is consistent with what clinicians have long assumed, but the dosage result gives the field something more specific to work with. It suggests the disagreement between trials is less about exercise style or delivery and more about how much exercise people actually receive. Because the analysis compared whole studies rather than individual patients, it cannot say which people are most likely to benefit, and that remains the obvious open question.

For someone with nonspecific chronic low back pain, this is a reason to think the specific style of exercise matters less than the amount accumulated, since dosage was the only feature linked to bigger benefits. It is also a reason to keep expectations realistic: the average improvements in pain and disability were small rather than transformative. And because the analysis worked at the level of whole studies, it says nothing about which individuals respond best.