Abstract

Background: The prognosis for functional recovery is crucial for discharge planning, optimizing healthcare resources and improving recovery chances. Prediction tools are key but must be usable from the beginning of rehabilitation and easy to use in clinical routine. Simple mobility milestones have shown predictive potential for stroke recovery. This study aims to identify the most efficient milestones for predicting the severity level of stroke patients' mobility at an early stage of recovery. Methods: In this retrospective cohort study medical records from a private hospital were reviewed through convenience sampling. The data collected included the Standardized Five Questions, Functional Status Score for the Intensive Care Unit, Berg Balance Scale, and Modified Rankin Scale, all unified by the Complexity-Based Model for Categorizing Rehabilitation, establishing severity levels. To predict mobility severity levels at reevaluations, CHAID (Chi-Square Automatic Interaction Detector) decision tree analysis and exact logistic regression analysis were employed. This involved using conventional mobility milestones alongside the clinical and demographic characteristics of the patients. Results: 50 patients met the criteria from admission (Evaluation 1: first week from stroke) to the first discharge point (Evaluation 2: second week from stroke). Only 25 patients remained for the second discharge point (Evaluation 3: fourth week from stroke). There were significant differences when comparing the functional assessments. The milestones “Sitting to standing” and “Sitting unsupported” were the most efficient predictors of future mobility complexity. Conclusions: “Sitting balance” and “Sitting unsupported” were the best predictors. These simple milestones are feasible to assess in daily clinical practice.