ABSTRACT
BACKGROUND: Ipsilateral hand (ILH) impairment is documented following motor stroke, but its impact on long-term outcome remains unknown. We assessed ILH impairment in subacute stroke and tested whether ILH impairment predicted long-term outcome. METHODS: We performed a longitudinal study in 209 consecutive patients with unilateral stroke and sensorimotor deficit at admission. ILH impairment was evaluated using the Purdue Pegboard Test (PPT) and handgrip strength and defined as mild (z-score < -1) or moderate (z-score < -1.65). We used logistic regression (LR) to predict outcome assessed 9 (range, 7-12) months post-stroke with the modified Rankin scale (mRS) categorized into good (mRS ≤ 1) and poor outcome (mRS ≥ 2). For internal validation, LR-bootstrapping and cross-validation with LASSO and Random Forest were performed. RESULTS: ILH impairment assessed at 89.04 ± 45.82 days post-stroke was moderate in 10.53% (95% CI 6.7, 14.83) for PPT and 17.22% (95% CI 11.96, 22.49) for grip, and mild in 21.05% (95% CI 15.78, 26.79) for PPT and 35.89 (95% CI 29.67, 42.58) for grip. Good outcome was predicted by ILH-PPT (B = 1.03 [95% CI 0.39, 3.31]), ILH-grip (B = 1.16 [95% CI 0.54, 3.53]), low NIHSS-discharge (B = -1.57 [95% CI -4.0, -1.19]), and no depression (B = -0.62 [95% CI -1.63, -0.43]), accounting for stroke delay (B = -0.011 [95% CI -0.06, 0.01]). Model efficiency was 91.6% (AUC = 0.977; 95% CI 0.959, 0.996). LASSO and Random Forest methods provided similar results, confirming the LR model robustness. CONCLUSIONS: ILH impairment is frequent after motor stroke and predicts long-term outcome. We propose to integrate ILH impairment into rehabilitation programs to improve recovery and serve research interventions such as neuromodulation.
