Comparative efficacy and acceptability of non-invasive neuromodulation technologies and botulinum toxin injections for post-stroke spasticity and motor function: a network meta-analysis of randomised controlled trials.

Background: Non-invasive neuromodulation is a promising approach for improving spasticity and motor function after stroke. However, it is still unclear which type of non-invasive neuromodulation is effective and evidence of important differences between them and botulinum toxin (BoNT) injection is limited. We aimed to assess the comparative efficacy and acceptability of non-invasive neuromodulation technologies and BoNT for post-stroke spasticity and motor function. Methods: In this network meta-analysis, Cochrane Library, EMBASE, MEDLINE, Web of Science, Scopus, CNKI, and Wan Fang Data were searched from the earliest records to 8 October 2024. Randomised controlled trials that compared any type of non-invasive neuromodulation therapies, BoNT, and control treatments (including sham or no stimulation/injection) for post-stroke spasticity measured by modified Ashworth scale (MAS) were included. MAS, motor function, and acceptability were pooled using random-effects model with summary weighted mean difference (WMD) or risk ratios (RR) alongside 95% confidence interval (CI). Ranking probabilities of the treatments were estimated. Clinical importance was categorized as definite, probable, possible, or definitely not, considering the relationship between effect measures (95% CI) and minimal clinically important difference (1, 6, and 1.5 points for MAS, motor function, and acceptability, respectively). The quality of evidence was assessed using CINeMA online web. PROSPERO registration CRD42024543494. Findings: 6260 studies were identified and 185 trials (11,185 participants; 12 interventions) were included. Compared with control treatments, BoNT, high- and low-frequency repetitive transcranial magnetic stimulation (HFrTMS and LFrTMS), and anodal, cathodal, and dual transcranial direct current stimulation (atDCS, ctDCS, and dtDCS) significantly improved spasticity at short-term follow-up (WMD range -0.81 to -0.31), but did not achieve clinical importance. At mid-term, ctDCS (WMD = -2.00; 95% CI: -3.03, -0.97) and dtDCS (WMD = -1.62; 95% CI: -3.22, -0.02) were more efficacious than control treatments in reducing post-stroke spasticity with probable clinical importance. For motor function, atDCS, ctDCS, and dtDCS were more efficacious than control treatments (WMD range 6.29-13.00), with probable clinical importance, while BoNT, HFrTMS, and LFrTMS with possible clinical importance (WMD range 3.42-5.28). Various modalities have comparable acceptability to control treatments (RR range 0.48-1.46). Confidence in accordance with CINeMA ranged from high to low. Sensitivity and meta-regression analyses on limb measured, cointervention, and stroke stage confirmed the main findings of this study. Interpretation: Taken together with clinical importance, evidence available supports three forms of tDCS as effective treatments for post-stroke spasticity and/or motor impairments, whereas BoNT, HFrTMS, and LFrTMS for motor impairments. These modalities could be considered alongside rehabilitation interventions as core treatments for post-stroke spasticity and motor impairments. Funding: China Postdoctoral Science Foundation (2024M752230).