Exercise and Chronic Low Back Pain — What Does the Cochrane Review Show?

Low back pain is one of the most common complaints that bring people to a doctor — in most cases it's non-specific, meaning it can't be attributed to a single, specific, serious structural cause visible on imaging. In some people an acute episode of pain resolves within a few weeks, but in a significant proportion of patients the pain persists for longer than 12 weeks — at that point we call it chronic pain, which follows different treatment rules than a fresh, acute episode.

For decades, the intuitive and widely recommended response to back pain was rest — lying down, avoiding movement, waiting for it to "go away on its own." This intuition has its own logic: if movement hurts, avoiding movement seems reasonable. The problem is that with chronic pain, prolonged immobilization has the opposite effect of what's intended — it weakens the muscles that stabilize the spine, worsens fitness, and in many cases increases, rather than eases, the discomfort felt over the long term.

The most reliable single source of knowledge about the effectiveness of exercise for chronic low back pain is the Cochrane systematic review — an organization known for rigorous methodology and regular updates to its conclusions as new studies emerge. The 2021 update covered a huge number of clinical trials, making it one of the most comprehensive summaries of the topic currently available.

In other words: exercise produces a clear, clinically noticeable difference in pain intensity compared with doing nothing — but the effect on everyday functioning is real, though more modest, and the difference compared with other active conservative treatment methods is even smaller still. This is an important distinction: exercise isn't a miracle cure that eliminates pain, but it consistently outperforms passive waiting.

This shift in approach doesn't mean rest is always bad — a short, few-day rest can be justified for a fresh, acute episode of pain. The problem arises when rest is prolonged and becomes the main, long-term "treatment plan" for chronic pain, instead of being a brief transitional stage before a gradual return to activity.

One of the more interesting observations from this review is how large a share of the included studies didn't compare exercise with no exercise at all, but instead compared different types of exercise with each other — as many as 61% of studies (151 of 249) tested two or more different forms of training, and 57% (142 studies) compared exercise with a non-exercise method. The sheer scale of these comparisons shows that the question of "which type of exercise is best" is studied intensively in the literature, but there is no single clear winner.

Instead, the review shows a consistent pattern from another angle: exercise as a category fared favorably against most of the alternatives it was compared with — it was probably more effective than patient education alone or electrotherapy — but it didn't differ significantly from manual therapy (a difference of 1.0 point, 95% CI -3.1 to 5.1 — practically no difference). This suggests that the mere fact of incorporating active, targeted movement matters more than choosing one specific method among many comparably effective options.

The review also compared exercise with specific alternative forms of conservative therapy. Compared with education or advice alone, without a targeted movement program, exercise produced a pain improvement of 12.2 points (95% CI -19.4 to -5.0) — a difference judged probably clinically significant. Compared with physiotherapy not involving exercise (e.g., physical treatments without an active movement component), the difference was 10.4 points (95% CI -15.2 to -5.6). The only comparison where no significant difference was found was against manual therapy.

The review authors noted that the certainty of evidence for pain outcomes was downgraded mainly because of substantial heterogeneity between studies — different populations, exercise protocols, and pain-measurement methods make it difficult to derive one precise number describing the effect. For functional ability outcomes, certainty was downgraded mainly because of some signs of publication bias (a tendency to publish studies with positive results more often). These are standard methodological caveats for such large pooled analyses, not a reason to dismiss the conclusions — Cochrane still rates the certainty of evidence as moderate, not low.

Of the studies included in the review, 86 reported any adverse effects related to therapy. In exercise groups, they were reported in 37 of 112 groups (33%), and in comparison groups in 12 of 42 groups (29%) — a small difference, and the adverse effects themselves were mostly mild, such as muscle soreness after training. Among the 12 studies that measured adverse effects systematically, the median number of events per participant was 0.14 (IQR 0.01–0.57) in exercise groups and 0.12 (IQR 0.02–0.32) in comparison groups.

Similar caution is warranted for pain accompanied by unintentional weight loss, fever, a history of cancer, or pain that worsens at night and doesn't depend on body position — these are so-called "red flag" symptoms, which should prompt a medical consultation rather than starting a training program on your own based on this or any other article.

It's rare in medicine to see such a clear reversal of a recommendation — from "rest until it passes" to "move, because it helps" — backed by a review covering nearly 250 clinical trials. It's precisely the scale and consistency of this data, not a single, promising study, that makes this conclusion a credible foundation for practical recommendations for people with chronic low back pain.