Therapeutic ultrasound for pain in knee osteoarthritis: a multilevel meta-analysis of treatment parameters

Background Therapeutic ultrasound is widely used for pain relief, but the extent to which treatment parameters are associated with clinical outcomes remains insufficiently characterized. This study aimed to evaluate the analgesic effect of therapeutic ultrasound in knee osteoarthritis and explore study-level associations between treatment parameters and effect size. Methods PubMed, Embase, MEDLINE, Web of Science, and the Cochrane Library (CENTRAL) were searched from inception to September 5, 2026. Randomized controlled trials evaluating therapeutic ultrasound for pain in adults with primary knee osteoarthritis were included. Effect sizes were expressed as Hedges’ g and synthesized using a multilevel random-effects meta-analysis. Study-level cluster-robust sensitivity analysis, univariable meta-regression with Benjamini–Hochberg false discovery rate (BH-FDR) correction, subgroup analyses, risk-of-bias assessment, and GRADE evaluation were performed. Results Twenty-three RCTs comprising 1436 participants and 55 effect sizes were included. Therapeutic ultrasound was associated with a moderate reduction in pain (Hedges’ g = 0.540, 95% CI 0.311–0.769; p < .001), although substantial heterogeneity was present and the 95% prediction interval crossed the null (−0.628 to 1.708). The study-level cluster-robust sensitivity analysis yielded a comparable estimate (95% CI 0.304–0.775). In the overall meta-regression, total session number remained significant after BH-FDR correction (adjusted p = .027), but the association was no longer significant after exclusion of an exceptionally prolonged treatment protocol ( p = .359). In exploratory modality-specific analyses, most nominal associations did not remain significant after correction. Only treatment duration in weeks (adjusted p = .005) and total session number (adjusted p = .007) in the low-intensity pulsed ultrasound subgroup remained clearly significant, although both associations were based on only six study clusters. No significant between-subgroup differences were observed according to ultrasound modality, osteoarthritis severity, or add-on treatment. The certainty of evidence was rated as moderate. Conclusions Therapeutic ultrasound was associated with a moderate average reduction in knee osteoarthritis pain, but substantial heterogeneity limits the generalizability of the pooled effect. Current evidence does not establish a superior ultrasound modality or a robust treatment-parameter relationship. Exploratory associations require confirmation in adequately powered, parameter-controlled randomized trials. Trial registration PROSPERO CRD420251175237.