Trajectories of arm recovery early after stroke: an exploratory study using latent class growth analysis

Abstract

Aim: To investigate trajectories of recovery of motor arm function after stroke during inpatient rehabilitation. Materials and methods: Data were available from 74 consecutively-admitted stroke survivors receiving inpatient rehabilitation from an inception cohort study. Heterogeneity of arm recovery in the first 4-weeks was investigated using latent class analysis and weekly Box and Block Test (BBT) scores. Optimal number of clusters were determined; characterised and cluster associated factors explored. Results: A 4-cluster model was identified, including 19 participants with low baseline arm function and minimal recovery (‘LOWstart/LOWprogress’, 26%), 15 with moderate function and low recovery (‘MODstart/LOWprogress’, 20%), 15 with low function and high recovery (‘LOWstart/HIGHprogress’, 20%), and 25 with moderate function and recovery (‘MODstart/MODprogress’, 34%). Compared to LOWstart/LOWprogress: LOWstart/HIGHprogress presented earlier post-stroke (β, 95%CI) (−4.81 days, −8.94 to −0.69); MODstart/MODprogress had lower modified Rankin Scale scores (−0.74, −1.15 to −0.32); and MODstart/LOWprogress, LOWstart/HIGHprogress and MODstart/MODprogress had higher admission BBT (23.58, 18.82 to 28.34; 4.85, 0.85 to 9.61; 28.02, 23.82 to 32.21), Upper Limb-Motor Assessment Scale (9.60, 7.24 to 11.97; 3.34, 0.97 to 5.70; 10.86, 8.77 to 12.94), Action Research Arm Test (31.09, 22.86 to 39.33; 12.69, 4.46 to 20.93; 38.01, 30.76 to 45.27), and Manual Muscle Test scores (10.64, 7.07 to 14.21; 6.24, 2.67 to 9.81; 11.87, 8.72 to 15.01). Conclusions: We found unique patterns of arm recovery with distinct characteristics for each cluster. Better understanding of patterns of arm recovery can guide future models and intervention development.KEY MESSAGES Arm recovery early after stroke follows four distinct trajectories that relate to time post stroke, initial stroke severity and baseline level of motor arm function. Identification of recovery patterns gives insight into the uniqueness of individual’s recovery. This study offers a novel approach on which to build and develop future models of arm recovery.