Efficacy of stability exercise-based treatment strategies for non-specific neck pain: a network meta-analysis.

Background: Non-specific neck pain (NSNP) is a common musculoskeletal disorder. Stability exercise is widely used in conservative management, but the relative efficacy of different exercise strategies remains unclear. This network meta-analysis evaluated stability exercise-based strategies for NSNP. Methods: PubMed, Embase, Cochrane Library, and Web of Science were searched from inception to November 24, 2025. Randomized controlled trials (RCTs) involving adults with NSNP were included. Risk of bias was assessed using RoB 2.0. A Bayesian network meta-analysis was conducted, and continuous outcomes were expressed as mean differences (MDs) with 95% credible intervals (95% CrIs). Global consistency and convergence were assessed using ΔDIC and PSRF, respectively. The certainty of evidence for key comparisons was evaluated using CINeMA. Results: Eighteen RCTs involving 950 participants were included; 10 studies were judged as low risk of bias, 4 as having some concerns, and 4 as high risk. Global consistency and convergence were acceptable (all ΔDIC < 5; PSRF = 1), with low-to-moderate heterogeneity across outcomes. For pain, several stability exercise strategies significantly reduced visual analog scale (VAS) scores compared with self-organized exercise (SOE), with shoulder resistance training combined with chest flexibility training (SRT+CFT) showing the greatest estimated reduction (MD = -8.16, 95% CrI: -13.04 to -3.27). For function, no intervention showed a significant advantage for the neck disability index (NDI), although neck resistance training combined with core resistance training (NRT+CRT) had the highest SUCRA value (MD = -27.43, 95% CrI: -58.28 to 3.90). For range of motion (ROM), neck flexibility training combined with NRT (NFT+NRT) had the highest SUCRA values for neck flexion, extension, bilateral lateral flexion, and bilateral rotation. The certainty of evidence was generally low or very low, with few moderate-certainty estimates. Conclusion: Stability exercise-based strategies showed outcome-specific effects in NSNP. NRT, SRT, CRT, NFT, and their combinations may be more beneficial for pain relief and ROM improvement, whereas no clear superiority was found for functional recovery. These findings should be interpreted cautiously because certainty was generally low or very low and several treatment nodes were supported by limited evidence. Systematic review registration: https://www.crd.york.ac.uk/PROSPERO/view/, identifier CRD420251243575.