Guidelines recommend exercise as first line care for the treatment of chronic low back pain. A Cochrane review summarised in a PEDro blog in November 2021 concluded that exercise therapy most likely reduces pain when compared to minimal treatment, but the impact of exercise therapy on functional limitations when compared to minimal treatment and on pain and functional limitations when compared to other conservative care are probably small. However, this Cochrane review did not investigate the effects of different types of exercise therapy compared to non-exercise controls nor do any head-to-head comparisons of different exercise therapies. This network meta-analysis aimed to estimate the effects of different exercise therapies on pain and function compared to non-exercise controls and to other types of exercise in adults with chronic low back pain.
Guided by a prospectively registered protocol, sensitive searches were performed in seven databases (including Cochrane CENTRAL, Medline and PEDro) and two trial registries to identify randomised controlled trials evaluating exercise therapy for adults with low back pain (date of most recent search: 7 December 2020). The patients were adults with non-specific low back pain of at least 12 weeks duration. The intervention was any exercise therapy prescribed or planned by a health professional that involved conducting specific activities, postures and/or movements with a goal to improve low back pain outcomes. The type of exercise therapy was classified into 11 categories: general strengthening; stretching; core strengthening (or motor control); flexibility; aerobic; functional restoration; McKenzie therapy; Pilates; yoga; mixed; and, other. The comparators were non-exercise controls and a different category of exercise therapy. Non-exercise controls were classified as minimal treatment (no treatment, usual care, placebo, education, and ineffective interventions like electrotherapy) and other conservative treatment (psychological therapy, anti-inflammatory or analgesic medication, relaxation, manual therapy, physiotherapy not involving exercise, back school). The primary outcomes were pain intensity and functional limitations measured on any scale, and data were transformed to a 0-to-100-point scale (where 0 is no pain or functional limitation) for the analyses. A 15-point difference in pain and a 10-point difference in functional limitations were pre-specified to be clinically important. If outcomes were evaluated at multiple time points, data from the time point closest to 3 months post-randomisation were used in the primary analyses. Two independent reviewers selected trials for inclusion and evaluated trial quality, and disagreements were resolved by discussion or by arbitration from a third reviewer. Data were extracted by one reviewer and checked by at least one other reviewer. Trial quality was evaluated using the first version of the Cochrane risk of bias tool. The Confidence in Network Meta-Analysis (CINeMA) approach was used to evaluate certainty of evidence. Pair-wise meta-analyses and network meta-analysis was used to pool trials and calculate the mean between-group differences (and the associated 95% confidence intervals). The two non-exercise comparator categories (minimal treatment and other conservative treatment) and the 11 exercise categories were used as nodes in the network meta-analysis.
217 trials (20,969 participants) were included in the analyses. The average age of participants was 44 years, 56% were women and the average pain intensity at baseline was 51 on a 100-point scale. The 369 exercise groups were categorised as core strengthening (110 groups), mixed (96), general strengthening (44), aerobic (25), Pilates (24), stretching (17), other (15), yoga (13), functional restoration (10), McKenzie therapy (9), and flexibility (5). The 138 non-exercise comparison groups were categorised as minimal treatment (86 groups) and other conservative treatment (52).
All categories of exercise therapy reduced pain intensity (mean difference -19 to -7 points; low or moderate certainty) and functional limitations (-12 to -3; low to high) compared to minimal treatment. Pilates, McKenzie therapy, functional restoration, and core strengthening reduced pain (-11 to -6; low or moderate), and McKenzie therapy, flexibility, Pilates and functional restoration reduced functional limitation (-7 to -3; moderate) compared to other conservative treatment.
Some types of exercise were more effective than others.
- Pilates exercise reduced pain more than all other exercise categories (mean difference -12 to -4 points; low or moderate certainty) and reduced functional limitations more than stretching, other, aerobic, mixed, general strengthening, yoga, core strengthening and functional restoration (-7 to -3; moderate).
- McKenzie therapy reduced pain more than stretching, aerobic, flexibility, yoga, mixed, other and general strengthening (mean difference -8 to -4; moderate certainty), and reduced functional limitations more than stretching, other, aerobic, mixed, general strengthening, yoga, core strengthening and functional restoration (-8 to -4; moderate).
- Functional restoration exercise reduced pain more than stretching, flexibility, aerobic, yoga, mixed and other (mean difference -8 to -4; low or moderate certainty), and reduced functional limitations more than stretching, other, aerobic, mixed and general strengthening (-4 to -2; moderate).
- Core strengthening exercise reduced pain more than stretching, aerobic, flexibility, yoga and mixed (mean difference -6 to -5; low certainty), and moderate certainty evidence that core strengthening exercise reduced functional limitations more than stretching, other, aerobic and mixed (-3 to -2; low).
Pilates, McKenzie therapy, functional restoration and core strengthening exercise therapies were more effective than other types of exercise therapy for reducing pain intensity and functional limitations. Nevertheless, people with chronic low back pain should be encouraged to perform the exercise that they enjoy in order to promote adherence.
For readers who are not familiar with network meta-analysis, we recommend reading a research note on this topic published in the Journal of Physiotherapy .
Hayden JA, et al. Some types of exercise are more effective than others in people with chronic low back pain: a network meta-analysis. J Physiother 2021;67(4):252-62