Investigation of Minimal Clinically Important Difference of the Extended Barthel Index in Inpatients With Acute to Subacute Stroke in China.

AIM: To determine the minimal clinically important difference (MCID) of the Extended Barthel Index (EBI) in patients with stroke. METHODS: In total, 209 patients with stroke were enrolled from a rehabilitation hospital. The EBI was administered at baseline, day 1, and week 4 after enrolment. At week 4, occupational therapists, patients, and expense payers independently rated the perceived ability change in activities of daily living. The MCID was estimated using distribution-, receiver operating characteristic (ROC) curve-, anchor-, and regression-based methods. RESULTS: Distribution-based estimates produced small to moderate effect sizes. ROC curve analysis yielded optimal cutoffs around 2.5-3.5 points depending on the roles; sensitivity was high, whereas specificity was low. Anchor-based and regression methods produced separate thresholds for deterioration and improvement across stakeholders. Based on the anchor-based method, the recommended MCID for deterioration ranged from -2.0 to 0 points and for improvement from 4.0 to 6.0 points, varying by rater perspective. CONCLUSION: Anchor-based MCID values are recommended for clinical use to distinguish improvement from deterioration, directly reflecting the perspectives from patients, therapists, and payers. The established thresholds can guide therapists' goal setting, progress monitoring, and rehabilitation policy decisions.