INTRODUCTION Low back pain is the leading contributor to the global disability and healthcare burdena. An increasing number of systematic reviews consistently demonstrated that exercise training improves physical function in individuals with low back painb,c. However, information about the optimal exercise prescription variables (e.g. duration, frequency, intensity) including methodological quality of the available evidence are mostly lackingd. Sub-optimal exercise prescription may increase injury risks and reduce potential for improvement. Therefore, we aimed to examine reporting of exercise prescription variables, assess methodological quality and exercise prescription recommendations in systematic reviews and meta-analyses of low back pain disorders.
METHODS We conducted a prospectively registered umbrella review of systematic reviews (CRD42021287440) for musculoskeletal diagnosis and focused here on low back pain. Peer-reviewed, English language journal articles were searched from inception to 14th February 2023 using the electronic databases of PubMed, SPORTDiscus, Cochrane Database of Systematic Reviews, EMBASE, and CINAHL. We included systematic reviews and meta-analyses of randomized controlled trials that compared exercise training to non-pharmacological, non-surgical conservative treatment, placebo, usual care, true control or other active exercise interventions for altering physical function, pain, mental health, adverse effects and treatment adherence in adults (>18 years) with musculoskeletal pain. Data were extracted from the following group of reviews: (1) those that did not report any exercise prescription data, and (2) those that performed a quantitative analysis of the relationship between exercise prescription variables and outcomes. AMSTAR-2 was used to assess methodological quality.
RESULTS From 6757 records, 274 systematic reviews were included for musculoskeletal pain, of which 72 represented evidence for low back pain. Seven (9.7%) reviews did not report any exercise prescription data, while 11 (15.3%) quantitatively analyzed the relationship between prescription variables and the outcome(s). Exercise duration (N=65) and frequency (N=54) were the most commonly reported prescription variables. Of the 11 reviews, all assessed pain and physical function, but only one reviewed mental health. No reviews examined the impact of exercise dose on adherence rates or adverse events. Subgroup analysis and meta-regression analysis revealed differences in effect sizes across reviews of different exercises training types and dosages with limited evidence for on an optimal dose. The overall methodological quality was rated as 'low or critically low' in 93.1% of reviews due to a lack of information about the critical domains in most of these reviews. Overlap analysis revealed a corrected covered area of 1.6%, indicating a slight degree of overlap among the included primary studies.
DISCUSSION Our review highlights the lack of high methodological quality evidence for optimal exercise training prescription variables for individuals with low back pain. Most reviews assessed only one type of exercise training mode and evaluated dose as categorical rather than continuous data. Exercise training dose recommendations varied among included reviews when prescribing exercise training. It is vital that future research: (1) identify optimum exercise prescription variables for low back pain, for example, via dose-response (network) meta-analysis, (2) perform high-quality reviews per AMSTAR-2 criteria, and (3) include outcomes of mental health, adverse events and exercise adherence.