Cathy M Stinear1, Winston D Byblow2, Suzanne J Ackerley2, P Alan Barber2, Marie-Claire Smith2. 1. From the Department of Medicine (C.M.S., S.J.A., P.A.B., M.-C.S.), Centre for Brain Research (C.M.S., W.D.B., S.J.A., P.A.B., M.-C.S.), and Department of Exercise Sciences (W.D.B.), University of Auckland, New Zealand; and Neurology, Auckland District Health Board, New Zealand (P.A.B.). c.stinear@auckland.ac.nz. 2. From the Department of Medicine (C.M.S., S.J.A., P.A.B., M.-C.S.), Centre for Brain Research (C.M.S., W.D.B., S.J.A., P.A.B., M.-C.S.), and Department of Exercise Sciences (W.D.B.), University of Auckland, New Zealand; and Neurology, Auckland District Health Board, New Zealand (P.A.B.).
Abstract
BACKGROUND AND PURPOSE: Several clinical measures and biomarkers are associated with motor recovery after stroke, but none are used to guide rehabilitation for individual patients. The objective of this study was to evaluate the implementation of upper limb predictions in stroke rehabilitation, by combining clinical measures and biomarkers using the Predict Recovery Potential (PREP) algorithm. METHODS: Predictions were provided for patients in the implementation group (n=110) and withheld from the comparison group (n=82). Predictions guided rehabilitation therapy focus for patients in the implementation group. The effects of predictive information on clinical practice (length of stay, therapist confidence, therapy content, and dose) were evaluated. Clinical outcomes (upper limb function, impairment and use, independence, and quality of life) were measured 3 and 6 months poststroke. The primary clinical practice outcome was inpatient length of stay. The primary clinical outcome was Action Research Arm Test score 3 months poststroke. RESULTS: Length of stay was 1 week shorter for the implementation group (11 days; 95% confidence interval, 9-13 days) than the comparison group (17 days; 95% confidence interval, 14-21 days; P=0.001), controlling for upper limb impairment, age, sex, and comorbidities. Therapists were more confident (P=0.004) and modified therapy content according to predictions for the implementation group (P<0.05). The algorithm correctly predicted the primary clinical outcome for 80% of patients in both groups. There were no adverse effects of algorithm implementation on patient outcomes at 3 or 6 months poststroke. CONCLUSIONS: PREP algorithm predictions modify therapy content and increase rehabilitation efficiency after stroke without compromising clinical outcome. CLINICAL TRIAL REGISTRATION: URL: http://anzctr.org.au. Unique identifier: ACTRN12611000755932.
BACKGROUND AND PURPOSE: Several clinical measures and biomarkers are associated with motor recovery after stroke, but none are used to guide rehabilitation for individual patients. The objective of this study was to evaluate the implementation of upper limb predictions in stroke rehabilitation, by combining clinical measures and biomarkers using the Predict Recovery Potential (PREP) algorithm. METHODS: Predictions were provided for patients in the implementation group (n=110) and withheld from the comparison group (n=82). Predictions guided rehabilitation therapy focus for patients in the implementation group. The effects of predictive information on clinical practice (length of stay, therapist confidence, therapy content, and dose) were evaluated. Clinical outcomes (upper limb function, impairment and use, independence, and quality of life) were measured 3 and 6 months poststroke. The primary clinical practice outcome was inpatient length of stay. The primary clinical outcome was Action Research Arm Test score 3 months poststroke. RESULTS: Length of stay was 1 week shorter for the implementation group (11 days; 95% confidence interval, 9-13 days) than the comparison group (17 days; 95% confidence interval, 14-21 days; P=0.001), controlling for upper limb impairment, age, sex, and comorbidities. Therapists were more confident (P=0.004) and modified therapy content according to predictions for the implementation group (P<0.05). The algorithm correctly predicted the primary clinical outcome for 80% of patients in both groups. There were no adverse effects of algorithm implementation on patient outcomes at 3 or 6 months poststroke. CONCLUSIONS: PREP algorithm predictions modify therapy content and increase rehabilitation efficiency after stroke without compromising clinical outcome. CLINICAL TRIAL REGISTRATION: URL: http://anzctr.org.au. Unique identifier: ACTRN12611000755932.
Authors: Lara A Boyd; Kathryn S Hayward; Nick S Ward; Cathy M Stinear; Charlotte Rosso; Rebecca J Fisher; Alexandre R Carter; Alex P Leff; David A Copland; Leeanne M Carey; Leonardo G Cohen; D Michele Basso; Jane M Maguire; Steven C Cramer Journal: Int J Stroke Date: 2017-07 Impact factor: 5.266
Authors: David J Lin; Alison M Cloutier; Kimberly S Erler; Jessica M Cassidy; Samuel B Snider; Jessica Ranford; Kristin Parlman; Fabio Giatsidis; James F Burke; Lee H Schwamm; Seth P Finklestein; Leigh R Hochberg; Steven C Cramer Journal: Stroke Date: 2019-10-25 Impact factor: 7.914
Authors: Adam de Havenon; Laura Heitsch; Abimbola Sunmonu; Robynne Braun; Keith R Lohse; John W Cole; Eva Mistry; Arne Lindgren; Bradford B Worrall; Steven C Cramer Journal: Arch Phys Med Rehabil Date: 2021-11-20 Impact factor: 3.966
Authors: Anna Palumbo; Viswanath Aluru; Jessica Battaglia; Daniel Geller; Alan Turry; Marc Ross; Adrian Cristian; Caitlin Balagula; Gbenga Ogedegbe; Latika Khatri; Moses V Chao; Robert C Froemke; Jacek K Urbanek; Preeti Raghavan Journal: Am J Phys Med Rehabil Date: 2021-12-06 Impact factor: 3.412
Authors: Alkis M Hadjiosif; Meret Branscheidt; Manuel A Anaya; Keith D Runnalls; Jennifer Keller; Amy J Bastian; Pablo A Celnik; John W Krakauer Journal: J Neurophysiol Date: 2022-02-02 Impact factor: 2.714