Identifying profiles of stroke patients benefitting from additional training: a latent class analysis approach

Objective: To identify profiles of stroke patient benefitting from additional training, using latent class analysis. Design: Retrospective observational study. Patients: Patients with stroke (n = 6,875) admitted to 42 recovery rehabilitation units in Japan between January 2005 and March 2016 who were registered in the Japan Association of Rehabilitation Database. Methods: The main outcome measure was the difference in Functional Independence Measure (FIM) scores between admission and discharge (referred to as “gain”). The effect of additional training, categorized as usual care (no additional training), self-exercise, training with hospital staff, or both exercise (combining self-exercise and training with hospital staff), was assessed through multiple regression analyses of latent classes. Results: Applying inclusion and exclusion criteria, 1185 patients were classified into 7 latent classes based on their admission characteristics (class size n = 82 (7%) to n = 226 (19%)). Patients with class 2 characteristics (right hemiparesis and modified dependence in the motor-FIM and cognitive-FIM) had positive FIM gain with additional training (95% confidence interval (95% CI) 0.49–3.29; p