Abstract
Objectives: Motor imagery (MI) may enhance post-stroke recovery, but evidence of its benefit over conventional rehabilitation therapy (CRT) is inconsistent. This study evaluated the effect of MI combined with CRT on upper-limb recovery, accounting for methodological quality and publication bias. Methods: A systematic review and meta-analysis was conducted following PRISMA guidelines. Searches were performed in multiple databases up to July 2025. Methodological quality and risk of bias were assessed using the PEDro scale and Cochrane RoB 2 tool, respectively. Analyses included the calculation of effect sizes (ES), heterogeneity, sensitivity, publication bias, and GRADE-based certainty assessment. Results: From 4074 records, 10 randomized controlled trials (n = 255) were included. The initial pooled analysis showed a small-to-moderate effect of MI + CRT versus CRT alone (ES = 0.45; 95% CI: 0.16-0.74). However, the overall ES calculated with a robust variance estimator was -0.06 (95% CI: -0.21, 0.08). Most trials had methodological limitations (mean PEDro = 6.0; high risk of bias in 7/10 studies). The GRADE evaluation indicated a very low certainty of evidence. Conclusions: The initially observed positive effect of MI combined with CRT is not robust. When accounting for statistical dependencies and potential biases, the effect vanishes and is no different from zero. Current evidence does not support the use of MI as a standalone adjunct to CRT. Larger, high-quality RCTs with standardized protocols are required to establish any potential clinical relevance.