Bayesian NMA dataset.

Low back pain (LBP) is the most common and incapacitating musculoskeletal disorder globally. The use of non-pharmacological interventions for the management of LBP is recommended by global clinical practice guidelines. However, existing guidelines often exhibit poor quality, as evidenced by inadequate systematic reviews of evidence. This necessitates further reviews tailored to the specific context of different countries. This review searched the CINAHL, EMBASE, MEDLINE, The Cochrane Library, Web of Science, and Wanfang databases from their inception to May 14, 2024. After data extraction by two independent researchers, a total of 57 randomized control trials (RCTs) were included. The Bayesian network meta-analysis results demonstrated that: (1) non-pharmacological therapies generally exhibited superior efficacy over pharmaceuticals in improving functional disability and overall efficacy; (2) pharmaceuticals, both alone and in combination with non-pharmacological therapies, were generally more effective than most non-pharmacological therapies in reducing pain intensity. High heterogeneity was observed, which could be explained by LBP subtypes in the analysis of functional disability. While heterogeneity had a limited impact on the confidence of results for pain intensity and functional disability, it significantly influenced the assessment of overall efficacy with major concerns of imprecision. The high volume of studies with a high risk of overall bias necessitates cautious interpretation of these findings. Chinese LBP patients may benefit most from non-pharmacological interventions, particularly those rooted in Traditional Chinese Medicine, for improving disability. For pain intensity reduction, pharmaceuticals and multi-component therapies incorporating pharmaceuticals may be more effective.