Abstract 112: Predicting REcovery Potential of Upper Limb Function After Stroke to Increase Rehabilitation Efficiency

Abstract

Listen

Translate article

Introduction: The PREP algorithm combines clinical assessment [Shoulder Abduction Finger Extension (SAFE) score], transcranial magnetic stimulation (TMS) and diffusion-tensor imaging to predict potential for upper limb recovery following stroke. Patients’ recovery potential is predicted to be Complete, Notable, Limited or None. Hypothesis: The PREP algorithm may be used in a ‘real world’ clinical setting to set individual rehabilitation goals. Methods: This study recruited 194 patients with upper limb weakness within 3 days of stroke. Assessments were made at baseline and 12 weeks by assessors blinded to PREP algorithm prediction. The initial benchmarking phase recruited 85 patients and PREP algorithm information was not shared with clinical teams or patients. The results were used to refine the algorithm and guide implementation in three ways. First, patients with a SAFE score > 7, predicted to have Complete upper limb recovery, were given a self-directed therapy program. Second, patients with a SAFE score of 5-7 could be given a Notable recovery prognosis, without requiring TMS. Third, 19% of patients exceeded their predicted upper limb recovery, so this possibility was conveyed to patients and clinical teams. The implementation phase recruited 109 patients, and PREP algorithm predictions were shared with patients and clinical teams. Results: Interim analyses (n = 135) find that the PREP algorithm correctly predicted upper limb function at 12 weeks for 85% of patients. Implementation of the algorithm decreased length of stay by 7 days (95%CI 2 - 15 days, p < 0.05) and increased the proportion of patients discharged home from the acute stroke unit from 28% to 49% (p < 0.01). Implementation also decreased upper limb therapy dose (p < 0.01), yet patient outcomes were similar between the two phases. Primary endpoint analysis will be complete in November 2015. Conclusions: Making predictions about the potential for recovery of upper limb function, and setting individual rehabilitation goals accordingly, may increase the efficiency of post-stroke rehabilitation.