Low back pain is a global health problem. Most international clinical practice guidelines recommend that first-line care for low back pain should involve non-pharmacological interventions, including those delivered by a physiotherapist. A recent review aimed to assess the quality and applicability of randomised controlled trials of physiotherapy interventions for low back pain.
The Physiotherapy Evidence Database (PEDro) was searched to obtain reports of trials evaluating physiotherapy interventions to prevent or treat low back pain (of any duration or type) in participants of any age. Data downloaded from PEDro were the citation (including year), participant age (paediatric <18 years; adult 18-70; geriatric >70), therapy codes, and PEDro scale (both total PEDro score and individual items) for methodology quality. Additional data extracted from the trials were type of research question (efficacy, effectiveness, economic evaluation, implementation or translation, unclear), intervention aim (prevention, treatment, combination), low back pain duration (acute <6 weeks, sub-acute 6-12, chronic >12, mixed-duration, not reported), and low back pain classification (non-specific, infection, fracture, inflammatory, radiculopathy, cancer, pregnancy, osteoporosis, mixed diagnosis, other). Two reviewers independently screened trials for inclusion and extracted the additional data, with disagreements resolved by discussion or arbitration by a third reviewer.
The analyses included 2,215 trials indexed in the 1 July 2019 update of PEDro. The majority of trials were for adults (n=2,136, 96%), low back pain without specific aetiology (n=1,863, 84%), and chronic duration (n=947, 43%). The top three most investigated therapy types were “stretching, mobilisation, manipulation, massage” (n=933, 42%), “strength training” (n=651, 29%), and “education” (n=499, 23%). The quality of trials improved over time, however most were at risk of bias. The mean total PEDro score was 5.4 (standard deviation 1.6) out of 10. Less than half of the trials concealed allocation to intervention (n=813, 37%), used intention-to-treat analysis (n=778, 35%), and blinded assessors (n=810, 37%), participants (n=174, 8%) and therapists (n=39, 2%). These findings did not vary by type of therapy.
The majority of trials evaluating physiotherapy interventions for low back pain are at risk of bias. Although average quality is improving with time, fundamental but simple to implement methodological features such as concealment of allocation and analysis by intention-to-treat are not commonly applied. Greater attention to these methodological features would improve the robustness of trials testing physiotherapy interventions for low back pain. The number of published trials relevant to low back pain has dramatically increased over time. Trials most commonly test exercise, education, and manual therapy interventions in adults with chronic, non-specific low back pain. Other interventions such as health promotion, and populations including children and older people are not well represented. Further efforts to improve the quality and applicability of the evidence are warranted.