Abstract: OBJECTIVE: Traditional meta-analyses have confirmed that exercise intervention can effectively improve chronic low back pain, but the evidence for which specific exercise method is more effective is still insufficient. Therefore, this article used a network meta-analysis to explore the differences in the effects of different exercise elements on improving chronic low back pain in adults. 
METHODS: Randomized controlled trials of exercise intervention for nonspecific chronic low back pain were searched in PubMed, Web of Science, Embase, Cochrane, CNKI, WanFang, and VIP databases. The trial group received any type of exercise intervention, while the control group received non-exercise interventions such as drug therapy and physical therapy. The search period was from the database inception to March 1, 2025. Boolean operators (AND/OR) were used to combine keywords for the search. Two reviewers independently completed literature screening, information extraction, and risk of bias assessment. A network meta-analysis was performed using Stata 17.0 software. The effects of exercise dosage variables were ranked using the Surface Under the Cumulative Ranking Curve (SUCRA). 
RESULTS: (1) A total of 40 randomized controlled trials were included. Network meta-analysis results showed that core stability training, mat exercises, traditional exercises, combined exercises, and other exercises were more effective than the control group in improving chronic low back pain [SMD=-0.76, 95% CI (-1.39, -0.13), P < 0.05; SMD=-1.67, 95% CI (-2.48, -0.86), P < 0.05; SMD=-2.09, 95% CI (-3.37, -0.80), P < 0.05; SMD=-1.60, 95% CI (-2.71, -0.49), P < 0.05]. 5; SMD=-1.40, 95% CI (-2.40, -0.40), P < 0.05]. Suspension training was less effective than traditional exercises and mat exercises in improving chronic low back pain [SMD=1.50, 95% CI (0.05, 2.95), P < 0.05; SMD=1.09, 95% CI (0.11, 2.06), P < 0.05]; traditional exercises were more effective than core stability training in improving chronic low back pain [SMD=-1.32, 95% CI (-2.64, -0.01), P < 0.05]. The intervention effect of single exercise sessions of 15-20 minutes and 30-40 minutes was better than that of the control group [SMD=-1.96, 95% CI (-3.55, -0.36), P < 0.05; SMD=-1.44, 95%CI (-2.12, -0.76), P < 0.05]. The intervention effects of exercising three times or six or seven times per week were better than those of the control group [SMD=-1.03, 95% CI (-1.69, -0.37), P < 0.05; SMD=-1.83, 95% CI (-2.75, -0.91), P < 0.05]. The intervention effect of exercising six or seven times per week was better than that of exercising once or twice per week [SMD=-1.30, 95% CI (-2.61, -0.06), P < 0.05]. The interventions of 4 weeks, 12-13 weeks and ≥16 weeks were significantly more effective than the control group [SMD=-0.81, 95% CI (-1.50, -0.12), P < 0.05; SMD=-1.63, 95%CI (-2.82, -0.43), P < 0.05; SMD=-2.14, 95% CI (-3.36, -0.92), P < 0.05]. The intervention of ≥16 weeks was significantly more effective than 6 weeks [SMD=-1.55, 95%CI (-3.03, -0.07), P < 0.05]. The SUCRA results showed that traditional exercise, 15-20 minutes per session, six or seven times per week, and ≥16 weeks of exercise ranked highest among their respective factors. 
CONCLUSION: Traditional exercises (Tai Chi, Qigong, Wuqinxi, and Baduanjin), intervention duration of 15–20 minutes, 6–7 sessions per week, and an intervention cycle of ≥16 weeks may offer the best efficacy for alleviating chronic low back pain in adults. However, due to the limited number of included studies, further research is needed to provide more robust evidence.

Key words: adults, chronic, chronic low back pain, exercise, network meta-analysis, exercise intervention