Time-dependent comparative efficacy of non-surgical treatments for pain relief in lateral epicondylitis: a systematic review and network meta-analysis

Background Lateral epicondylitis (LE) is a common tendinopathy, but the relative pain-relieving efficacy of different non-surgical interventions across follow-up stages remains unclear. This study used a Bayesian network meta-analysis to compare the time-dependent effects of multiple non-surgical treatments for pain relief in LE. Methods PubMed, Web of Science, and the Cochrane Central Register of Controlled Trials were systematically searched from database inception to March 5, 2024, with an updated search conducted to February 2, 2026. Randomized controlled trials involving adults with LE who received non-surgical interventions were included. Pain intensity measured by the visual analog scale (VAS) was the primary outcome, and all VAS scores were standardized to a 0–10 scale, with lower scores indicating less pain. Post-treatment VAS scores were synthesized in three predefined time windows: short-term (1–4 weeks; the result closest to 4 weeks), intermediate-term (4–12 weeks; the result closest to 12 weeks), and long-term (>12 weeks; the longest follow-up beyond 12 weeks). A Bayesian random-effects network meta-analysis was performed. Treatment effects were expressed as mean differences (MDs) with 95% credible intervals (CrIs), and ranking probabilities were summarized using the surface under the cumulative ranking curve (SUCRA). Results A total of 27 randomized controlled trials were included. In the short term, kinesio taping (KT), corticosteroid injection (CSI), brace, and laser therapy (LA) showed superior pain relief compared with placebo; the MDs (95%CrIs) for KT and CSI were -4.10 (-6.14 to -2.11) and -3.57 (-5.71 to -1.47), respectively. In the intermediate term, CSI, extracorporeal shock wave therapy (ESWT), glycosaminoglycan polysulfate (GAGPS), KT, physical therapy (PT), pulsed ultrasound (PU), and ultrasound (US) were superior to placebo; the MDs (95%CrIs) for KT and CSI were -2.58 (-3.92 to -1.32) and -1.60 (-2.77 to -0.44), respectively. In the long term, no intervention showed a statistically significant advantage over placebo. Although KT and CSI ranked relatively high in the short term, and GAGPS and KT ranked relatively high in the intermediate term, ranking results should be interpreted cautiously in light of interval width, direct evidence, and network consistency. Conclusion The pain-relieving effects of non-surgical interventions for LE appear to be time-dependent. Some treatments may be more favorable for short- or intermediate-term pain relief, but evidence for long-term superiority remains insufficient. Clinical interpretation should not rely on treatment ranking alone, but should instead integrate the follow-up stage, effect estimates, and evidence certainty. More high-quality randomized controlled trials with long-term follow-up are needed to clarify the long-term value of different non-surgical treatment strategies.