Effects of Navigated rTMS on Post-Stroke Upper-Limb Function: A Systematic Review and Meta-Analysis of Randomized...

Objectives: Neuronavigation may improve the precision and reproducibility of repetitive transcranial magnetic stimulation (rTMS) by aligning stimulation with individualized targets. Whether navigation-guided rTMS benefits post-stroke upper-limb recovery is unclear. We conducted a PRISMA-compliant systematic review and meta-analysis to estimate the effect of navigated rTMS, added to standard rehabilitation, versus sham. Methods: The protocol was registered in PROSPERO (CRD420251165052). Two reviewers independently searched CENTRAL, MEDLINE, Embase, CINAHL, Web of Science, and Google Scholar (October 2025), screened records, extracted data, and assessed risk of bias (Cochrane RoB-1). The prespecified primary endpoint was changed in Fugl-Meyer Assessment of the upper extremity (FMA-UE) from baseline to end of treatment. Effects were pooled as mean differences under random-effects models. When change-score standard deviations (SDs) were unavailable, they were derived from pre/post SDs assuming within-person correlation r = 0.5; sensitivity analyses used r = 0.7 and r = 0.9. Multi-arm trials were combined to avoid double counting. Results: four randomized, sham-controlled trials (n = 297) contributed end-of-treatment change in FMA-UE. The pooled effect favored navigated rTMS but was not statistically significant (MD 3.65, 95% CI -1.84 to 9.13; I² = 73%). Sensitivity analyses with higher r produced directionally consistent estimates. A subgroup of 2-week (10-session) protocols (k = 3) showed a significant benefit (MD 7.09, 95% CI 4.14 to 10.05; I² = 0%). Most risk-of-bias domains were low risk. Conclusions: Navigated rTMS did not show a consistent short-term advantage over sham on FMA-UE across heterogeneous protocols. A positive signal in standardized 2-week courses supports further adequately powered multicenter randomized controlled trials (RCTs) with harmonized protocols and complete variance reporting.